It's sad to see that ex-GitHub CEO didn't make enough money to just kick-start his company, but needs external money which will later on dictate how the company works or will sell users and the product for the next exit..
Every 'platform for AI agents' announcement makes me wonder: are we building tools for a workflow that actually exists, or are we building tools and hoping the workflow materializes? The GitHub parallel is instructive because GitHub succeeded by meeting developers where they already were (git). The question for Entire is whether agents have a natural coordination layer yet or if this is premature infrastructure.
I don't think your Github example is accurate. The vast majority of developers started using git after Github became a thing. They may have used svn or another type of collaboration system before, but not git. And the main reason they started using git is because Github was such massive value on top of git, not because git was so amazing.
I do not think that's how it worked out for GitHub: I'd rather say that Git (as complex as it was to use) succeeded due to becoming the basis of GitHub (with simple, clean interface).
At the time, there were multiple code hosting platforms like Sourceforge, FSF Savannah, Canonical's Launchpad.net, and most development was still done in SVN, with Git, Bazaar, Mercurial the upstart "distributed" VCSes with similar penetration.
Yes, development was being done in SVN but it was a huge pain. Continuous communication was required with the server (history lookups took ages, changing a file required a checkout, etc.) and that was just horribly inefficient for distributed teams. Even within Europe, much more so when cross-continent.
A DVCS was definitely required. And I would say git won out due to Linus inventing and then backing it, not because of a platform that would serve it.
Yes to all that. And GitLab the company was only founded in 2014 (OSS project started in 2011) and ran through YC in 2015, seven years after GitHub launched.
Of the thousands, a handful will prevail. Most of it is vaporware, just like in any boom. Every single industry has this problem; copy-cats, fakes & frauds.
"Buy my fancy oil for your coal shovel and the coal will turn into gold. If you pay for premium, you don't have to shovel yourself."
If everything goes right, there won't be a coal mine needed.
I'd bet that less people had their source code on git in 2008 than the number of developers using the various coding agents today. And the open-source project that we published today hooks into the existing workflow for those developers, in Claude Code and in Gemini CLI. Time will tell the rest. We will publish regular updates and you can judge us on those results.
At least for me, I have felt like the chat history in an agent is often times just as important and potentially even more important than the source code it generates. The code is merely the compiled result of my explanations of intent and goals. That is, the business logic and domain expertise is trapped in my brain, which isn't very scalable.
Versioning and tracking the true source code, my thoughts, or even the thoughts of other agents and their findings, seems like a logical next step. A hosted central place for it and the infrastructure required to store the immense data created by constantly churning agents that arrive at a certain result seems like the challenge many seem to be missing here.
We are building tools and hoping an exit materializes. There’s so much funny money in AI right now, getting life-altering money seems easily attainable
HN is full of AI agents hype posts. I am yet to see legitimate and functional agent orchestration solving real problems, whether it is scale or velocity.
This is the point of that post and helpfully it was added at the top in a TL;dr and was half of that t sentence TL;dr. Will succeed or not? Well, that's a coin toss, always been.
I mean, pretty much all big startups begin as "niche" things that people might care about later. Tesla, Airbnb, Twitch... and countless failures too. It's just how the game is.
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more.
This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
I currently develop small utilities with the help of AI, but am far from vibe coding or using agents. I review every single suggestion and do some refactoring at each step, before any commit (sometimes heavy refactoring; sometimes reorganizing everything).
In my experience LLMs tend to touch everything all of the time and don't naturally think about simplification, centralization and separation of concerns. They don't care about structure, they're all over the place. One needs to breathe on their shoulders to produce anything organized.
Maybe there's a way to give them more autonomy by writing the whole program in pseudo-code with just function signatures and let them flesh it out. I haven't tried that yet but it may be interesting.
What kind of barrier/moat/network effects/etc would prevent someone with a Claude Code subscription from replicating whatever "innovation" is so uniquely valuable here?
It's somewhat strange to regularly read HN threads confidently asserting that the cost of software is trending towards zero and software engineering as a profession is dead, but also that an AI dev tool that basically hooks onto Git/Claude Code/terminal session history is worth multiples of $60+ million dollars.
I built a basic copy with about an hour with my own "platform for ai agents" that I built out over the last week: https://github.com/jwbron/egg/pull/504, and refined it here: https://github.com/jwbron/egg/pull/517 (though right after I merged this I blew through my weekly token quota for my second claude max 20x account so I haven't been able to test it out yet).
I think your point is valid and I've been having the same thoughts. My tooling is still in the experimental phase, but I can move so quickly that I'm having trouble grasping how products like this will survive. If I can build this out in a week and copy an idea like this one (which is a great one, mind you) in an hour, what's the value of paying someone for a product like this vs just building it myself?
There's no way this company is just a few git and claude hooks with a CLI. They're definitely working on a SASS - something else that isn't open source that this primitive is the basis of. Like a GitHub for agent code
> What kind of barrier/moat/network effects/etc would prevent someone with a Claude Code subscription from replicating whatever "innovation" is so uniquely valuable here?
You are correct, that isn't the moat. Writing the software is the easy part
git didn't succeed based on the mental model. It got a foot in the door with better tooling and developer experience then blew the door open when GitHub found a way to productize it.
I have never seen any thread that unanimously asserts this. Even if they do, having HN/reddit asserting something as evidence is wrong way to look at things.
HN is full of AI agents hype posts. I am yet to see legitimate and functional agent orchestration solving real problems, whether it is scale or velocity.
For the last three or four months, what I've been doing is anytime I have Claude write a comment on an issue, it just adds a session ID, file path and the VM it is on. That way, whenever we have some stuff that comes up, we just search through issues and then we can also retrace the session that produced the work and it's all traceable. In general, I just work through gitea issues and sometimes beads. I couldn't stand having all these MD files in my repo because I was just drowning in documentation, so having it in issues has been working really nicely and agents know how to work with issues. I did have it write a gitea utility and they are pretty happy using/abusing it. Anytime I see that they call it in some way that generates errors, I just have them improve the utility. And by this point, it pretty much always works. It's been really nice.
predicting that a startup will fail is.. well, you got a ton of probability on your side there. so it isn't a particularly impressive thing to be right about.
Unimpressive doesn't mean incorrect, sometimes it's good to take the side of the most probable. And yet at the same time I am reminded of this quote:
> The reasonable man adapts himself to the world: the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. - George Bernard Shaw
I'm not disagreeing, just soliciting. Does anyone have examples of products that failed in the early stages because their implementation was too trivial?
I mean a dev tool that's seemingly failing to resonate with developers as to why they would pay for this is a pretty good way to tell if it's going to fall in the 1%.
The Dropbox take was wrong because they didn't understand the market for the product. This time the people commenting are the target audience. You even get the secondary way this product will lose even if turns out to be a good idea, existing git forges won't want to lose users and so will standardize and support attaching metadata to commits.
People keep saying that, but it's hardly the same thing. We're talking about developer workflow here. It's like someone coming up with Brancher. It's a git branch manager. Use `brancher foo` to replace `git checkout -b foo`. "Remember that comment about rsync and dropbox? Brancher is to git, what dropbox is to rsync"
How is LangChain doing? How about OpenAI's Swarm or their Agent SDK or whatever they called it? AWS' agent-orchestrator? The crap ton of Agent Frameworks that came out 8-12 months ago? Anyone using any of these things today? Some poor souls built stuff on it, and the smart ones moved away, and some are stuck figuring out how to do complex sub-agent orchestration and handoffs when all you need apparently is a bunch of markdown files.
Just saw a Discord-weekend take on reddit! Haha. Guy was saying he could create it in a day and then self-host it on his servers so that he doesn't have to put Nitro ads on top of it
It's the valuation that is wild to me (I think the idea itself has merit). But these are the new economics. I can only say "that's wild" enough before it is in fact no longer wild.
These aren't new economics, it's just VC funds trying to boost their holdings by saying it's worth X because they said so. Frankly the FTC should make it illegal.
That is where I’m shocked being in a position of raising for a startup myself, what was in their pitch deck/data room that convinced them of this valuation? Or is it due to the founders reputation and not the substance?
That's not impressive. That's an incredible amount concentrated in the hands of a few looking for a place to live. It has to end up somewhere. Some of it goes everywhere.
We have had this for ages now.... I just don't have access to the sort of people willing to pass me 60m for that. I never thought it to be worth anything really ; it was a trivial to implement afterthought.
I love this one so much! The arbitrary decision to cherry-pick critique a particular product to this degree, when it’s something that could be said about 99% of the stuff SV churns out, including in all likelihood anything you’ve ever worked on.
Good thing the comment you're replying to does not lionise 99% of the stuff SV churns out, including in all likelihood anything they've ever worked on. I guess we should just not critique anything out of SV because it's all shit?
Wow, read through the comments and you weren't joking. I attribute this to crossroads of "this release is v0.1 of what we are building" and the HN crowd who have been scrolling past 120 AI frameworks and hot takes daily and have no patience for anything that isn't immediately 100% useful to them in the moment.
I find the framing of the problem to be very accurate, which is very encouraging. People saying "I can roll my own in a weekend" might be right, but they don't have $60M in the bank, which makes all the difference.
My take is this product is getting released right now because they need the data to build on. The raw data is the thing, then they can crunch numbers and build some analysis to produce dynamic context, possibly using shared patterns across repos.
Despite what HN thinks, $60M doesn't just fall in your lap without a clear plan. The moat is the trust people will have to upload their data, not the code that runs it. I expect to see some interesting things from this in the coming months.
I haven't read the article yet but this conversation reminds me of Docker. Lots of people "didn't get it." I told them at the time: if you don't get it you aren't ready for it yet so don't worry about it. When you do need it, you'll get it and then you'll use it and never look back. Look at where we are with containers now.
100% agree because there’s a lot of value in understanding how and why past code was written. It can be used to make better decisions faster around code to write in the future.
E.g., if you’ve ever wondered why code was written in a particular way X instead of Y then you’ll have the context to understand whether X is still relevant or if Y can be adopted.
E.g., easier to prompt AI to write the next commit when it knows all the context behind the current/previous commit’s development process.
But that's not what is in the whole context. The whole context contains a lot of noise and false "thoughts".
What the AI needs to do is to document the software project in an efficient manner without duplication. That's not what this tool is doing.
I question the value in storing all the crap in git.
ehhhh is it really that useful though? Sounds way more noisy than anything, and a great way to burn through tokens. It's like founding a startup to solve the problem of people squashing their commits. Also, it sounds like something Claude Code/Codex/etc could quickly add an extension for.
Maybe use critical thinking instead of a mindless dismissal?
The fact that you aren't haven't offered a single counterargument to any other posters' points and have to resort to pearl-clutching is pretty good proof that you can't actually respond to any points and are just emotionally lashing out.
This is literally what claude code already does minus the commit attachment. It’s just very fancy marketing speak for the exact same thing.
I’m happy to believe maybe they’ll make something useful with 60M (quite a lot for a seed round though), but Maybe not get all lyrical about what they have now.
Please don't use quotation marks to make it look like you're quoting someone when you aren't. That's an internet snark trope and we're trying to avoid those on HN.
Look it’s obvious at this point to anyone who is actually using the tools.
We can articulate it but why should we bother when it’s so obvious.
We are at an inflection point where discussion about this, even on HN, is useless until the people in the conversation are on a similar level again. Until then we have a very large gap in a bimodal distribution, and it’s fruitless to talk to the other population.
Some Tom Dick and Harry to VCs: I have a proposal for you.
VCs: what is it
Tom Dick & Harry: AI
VCs: get the ** out of here, we already burnt enough money and will never see it back
Tom Dick & Harry: hear me out this is different
VCs: ok you have 5 minutes to explain me your product
Tom Dick & Harry: I dont have one
VCs: get the ** out of here
Tom Dick & Harry: hear me out
VCs: ok, you have 30 seconds to impress us.
Tom Dick & Harry: I just quit Microslop and still have high level contacts there
VCs: Hot damn!!! you are our lottery ticket to recoup all the money we have lost in other ventures. This is going to be a race against time, before your contacts go stale. Here's 60M for you, wine and dine your friends with it. On your way out you will find some AI generated product names and some vague product descriptions. Pick one and slap it on some website and announce our deal. Now get the ** out of here.
The thread is missing the forest for the trees. The interesting bet here isn't git checkpoints—it's that someone is finally building the observability layer for agent-generated code.
Most agent frameworks (LangChain, Swarm, etc.) obsessed over orchestration. But the actual pain point isn't "how do I chain prompts"—it's "what did the agent do, why, and how do I audit/reproduce it?"
The markdown-files-in-git crowd is right that simple approaches work. But they work at small scale. Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs.
The $60M question is whether that problem is big enough to justify a platform vs. teams bolting on their own logging. I'm skeptical—but the underlying insight (agent observability > agent orchestration) seems directionally correct.
@dang with the launch of open claw I have seen so much more LLM slop comments. I know meta comments like mine aren't usually encouraged, but I think we need to do something about this as a community. Is there anything we can do? (either ban or at least requiring full disclosure for bot comments would be nice).
EDIT: I suspect the current "solution" is to just downvote (which I do!), but I think people who don't chat with LLMs daily might not recognize their telltale signs so I often see them highly upvoted.
Maybe that means people want LLM comments here, but it severely changes the tone and vibe of this site and I would like to at least have the community make that choice consciously rather than just slowly slide into the slop era.
Parent comment has the rhythm of an AI comment. Caught myself not realizing it until you mentioned it. Seems like I am more in tune with LLM slop on twitter, which is usually much worse.
But on second sight it's clear and it also shows the comment as having no stance, and very generic.
@dang I would welcome a small secondary button that one can vote on to community-driven mark a comment as AI, just so we know.
Yes I usually just bite my tongue and downvote, but with the launch of open claw I think the amount of slop has increased dramatically and I think we need to deal with it sooner than later.
How much experience do you have interacting with LLM generated prose? The comment I replied to sets off so many red flags that I would be willing to stake a lot on it being completely LLM generated.
It's not just the em dashes - its the cadence, tone and structure of the whole comment.
Yeah it's really frustrating how often I see kneejerk rebuttals assuming others are solely basing it on presence of em-dashes. That's usually a secondary data point. The obvious tells are more often structure/cadence as you say and by far most importantly: a clear pattern of repeated similar "AI smell" comments in their history that make it 100% obvious.
> Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs.
This has been the story for every trend empowering developers since year dot. Look back and you can find exactly the same said about CD, public cloud, containers, the works. The 'orchestration' (read compliance) layers always get routed around. Always.
I thought everyone was just using open telemetry traces for this? This is just a classic observability problem that isn’t unique with agents. More important yes, but not unique functionally.
Ok, I’ll grant you that if they can get agents to somehow connect to other’s reasoning in realtime that would be useful. Right now it’s me that has to play reasoning container.
This is interesting. I’m experimenting with something adjacent in an open source plugin, but focused less on orchestration and more on decision quality.
Instead of just wiring agents together, I require stake and structured review around outputs. The idea is simple: coordination without cost trends toward noise.
Curious how entire.io thinks about incentives and failure modes as systems scale.
That is a sharp observation———it is the observability that matters! The question arises: Who observes the observers? Would you like me to create MetaEntire.ai———an agentic platform that observes Entire.io?
verbatim llm output with little substance to it.
HN mods don't want us to be negative but if this is what we have to take serious these days it is hard to say anything else.
I guess I could not comment at all but that feels like just letting the platform sink into the slopacolypse?
I have CURRENT_TASK.md that does more or less the same thing. It also gets committed to git. So I guess that’s entire? Wish I’d realized I was sitting on a 60M idea…
> Checkpoints run as a Git-aware CLI. On every commit generated by an agent, it writes a structured checkpoint object and associates it with the commit SHA. The code stays exactly the same, we just add context as first-class metadata. When you push your commit, Checkpoints also pushes this metadata to a separate branch (entire/checkpoints/v1), giving you a complete, append-only audit log inside your repository. As a result, every change can now be traced back not only to a diff, but to the reasoning that produced it.
The context for every single turn could in theory be nearly 1MB. Since this context is being stored in the repo and constantly changing, after a thousand turns, won't it make just doing a "git checkout" start to be really heavy?
For example, codex-cli stores every single context for a given session in a jsonl file (in .codex). I've easily got that file to hit 4 GB in size, just working for a few days; amusingly, codex-cli would then take many GB of RAM at startup. I ended up writing a script that trims the jsonl history automatically periodically. The latest codex-cli has an optional sqlite store for context state.
My guess is that by "context", Checkpoints doesn't actually mean the contents of the context window, but just distilled reasoning traces, which are more manageable... but still can be pretty large.
> won't it make just doing a "git checkout" start to be really heavy?
not really? doesn't git checkout only retrieve the current branch? the checkpoint data is in another branch.
we can presume that the tooling for this doesn't expect you to manage the checkpoint branch directly. each checkpoint object is associated with a commit sha (in your working branch, master or whatever). the tooling presumably would just make sure you have the checkpoints for the nearby (in history) commit sha's, and system prompt for the agent will help it do its thing.
i mean all that is trivial. not worth a $60MM investment.
i suspect what is really going on is that the context makes it back to the origin server. this allows _cloud_ agents, independent of your local claude session, to pick up the context. or for developer-to-developer handoff with full context. or to pick up context from a feature branch (as you switch across branches rapidly) later, easily. yes? you'll have to excuse me, i'm not well informed on how LLM coding agents actually work in that way (where the context is kept, how easy it is to pick it back up again). this is just a bit of opining based on why this is worth 20% of $300MM.
if i look at https://chunkhound.github.io it makes me think entire is a version of that. they'll add an MCP server and you won't have to think about it.
finally, because there is a commit sha association for each checkpoint, i would be worried that history rewrites or force pushes MUST use the tooling otherwise you'd end up screwing up the historical context badly.
This feels interesting because the real problem with AI generated code isn't that it is of better or worse quality than code written by a human, it's that we humans need to audit both types of code. And this platform feels like it addresses this in a novel and traditional way. I like it.
> Spec-driven development is becoming the primary driver of code generation.
This sounds like my current "phase" of AI coding. I have had so many project ideas for years that I can just spec out, everything I've thought about, all the little ideas and details, things I only had time to think about, never implement. I then feed it to Claude, and watch it meet my every specification, I can then test it, note any bugs, recompile and re-test. I can review the code, as you would a Junior you're mentoring, and have it rewrite it in a specific pattern.
Funnily enough, I love Beads, but did not like that it uses git hooks for the DB, and I can't tie tickets back to ticketing systems, so I've been building my own alternative, mine just syncs to and from github issues. I think this is probably overkill for whats been a solved thing: ticketing systems.
I am going lower level - every individual work item is a "task.md" file, starts initially as a user ask, then add planning, and then the agent checks gates "[ ]" on each subtask as it works through it. In the end the task files remain part of the project, documenting work done. I also keep an up to date mind map for the whole project to speed up start time.
And I use git hooks on the tool event to print the current open gate (subtask) from task.md so the agent never deviates from the plan, this is important if you use yolo mode. It might be an original technique I never heard anyone using it. A stickie note in the tool response, printed by a hook, that highlights the current task and where is the current task.md located. I have seen stretches of 10 or 15 minutes of good work done this way with no user intervention. Like a "Markdown Turing Machine".
That's hilarious, I called it gates too for my reimplementation of Beads. Still working on it a bit, but this is the one I built out a month back, got it into git a week ago.
For me a gate is: a dependency that must pass before a task is closed. It could be human verification, unit testing, or even "can I curl this?" "can I build this?" and gates can be re-used, but every task MUST have one gate.
My issue with git hooks integration at that level is and I know this sounds crazy, but not everyone is using git. I run into legacy projects, or maybe its still greenfield as heck, and all you have is a POC zip file your manager emailed you for whatever awful reason. I like my tooling to be agnostic to models and external tooling so it can easily integrate everywhere.
Yours sounds pretty awesome for what its worth, just not for me, wish you the best of luck.
Me too. I've been using spec-kitty [0], a fork of Spec Kit. Quite amazing how a short interview on an idea can produce full documents of requirements, specs, tasks, etc. After a few AI projects, this is my first time using spec driven development, and it is definitely an improvement.
Task management is fundamentally straightforward and yet workflow specific enough that I recommend everyone just spend a few hours building their own tools at this point.
I started off with the original beads and it was definitely a nightmare. However I would recommend using https://github.com/Dicklesworthstone/beads_rust - it's a much simpler implementation of the same concept, without all the random extra stuff thrown on to support Gas Town.
Ignoring the VC economics and awful name, I won’t be as pessimistic as everyone. I see the vision.
That said, nobody knows what the AI future looks like. Entire’s entire thesis is a solution for something we don’t even know we need. It’s a massive bet and uphill battle. Traditionally, dev tool success stories come from grassroots projects of developers solving their own problems and not massive VC funded efforts that tell you what you need to do.
My pessimism is mostly rooted in the VC economics of it all. The vision is great, but its a busy space and there's no actual product or business. They basically wrote the guy a check to build the spaceship in space.
Exactly ... tired by all the marketing hyperbole talk. Just show what your product does in a simple example / showcase. If it's good, people will like it. You can save yourself a lot of text copy and user time that way.
CEOs have many audiences; great CEOs communicate capably with each.
FWIW it's not entirely clear to me who Entire's long-term customer is, but the (interesting!) CLI that shipped today is very much for developers who are busy building with agents.
The problem is that when it comes to (commercial) developer tools and services, everyone can/wants to be everything, so why let a simple statement or a showcase limit you? "Hey, we are a container scanning service... But we can also be a container registry too, a CI, a KeyValue store, an agent sandbox provider, git hosting? We can do quick dev deployments/preview too. Want a private npm registry? Automated pull request reviews? Code Signing service? We are working on a new text editor btw"
I feel like these types of pages are less geared towards actual users of the product and more towards the investors who love the vague and flowery language. We're no longer in a world where the path to profitability was the objective goal anyway, it makes sense to me that the marketing of software is becoming decreasingly detached from reality..
It's almost like an extension of the "if you're not paying for the product, you are the product" idea. If you're assessing a tool like this and the marketing isn't even trying to communicate to you, the user, what the product does, aren't you also kind of "the product" in this case too?
The domain expired a few days ago and was purchased by someone else and then changed. There's a recreation of the original here https://html5zombo.com/
Seems they install a Git hook or something that executes on commit and saves your chatbot logs associated with the commit hash. This is expected to somehow improve on the issue that people are synthesising much more code than they could read and understand, and make it easier to pass along a bigger context next time you query your chatbots, supposedly to stop them from repeating "mistakes" that have already wasted your time.
What it does? Imagine a multi line commit message.
Yes yes a Dropbox comment. But the problem here is 1 million people are doing the same thing. For this to be worth 60M seed I suspect they need to do something more than you can achieve by messing around locally."
"Claude build me a script in bash to implement a Ralph loop with a KV store tied to my git commits for agent memory."
I think the moat depends on how long it takes for an agent to ingest the entire commit history and product documentation into context on the fly. At the rate models are improving, seeing the reasoning chain of an outdated model that led to a commit that warrants post-hoc review (likely becuase of a bug) would mainly be useful for root cause analysis more than for insight into what to do next... but chances are the newer model would have been able to infer it from local context anyway.
It also creates a challenge with respect with the embedding model chosen and how future proof it turns out to be.
I landed on a similar vision last year. The more I thought about it, the moat felt fragile. GitHub or GitLab could build the same capabilities and become a natural extension of what teams already use. That said, it addresses a real problem, and the SDLC needs to evolve.
Agents can save their reasoning into markdown files, and commit those files to Git. Are "Checkpoints" just a marketing term for that, or there's more to it?
Claude Code already does this, you can access it with /resume, /rewind and /fork. I'd imagine building a version that saves in the repo instead of in the home folder would take very minimal effort.
Hmm the whole point of checkpoints seems to be to reduce token waste by saving repeat thinking work. But wouldn't trying to pull N checkpoints into context of the N+1 task be MUCH more expensive? It's at odds with the current practice of clearing context regularly to save on input tokens. Even subagents (which I think are the real superpower that Claude Code has over Gemini CLI for now) by their nature get spawned with fresh near-empty context.
Token costs aside, arguably fresh context is also better at problem solving. When it was just me coding by hand, I didn't save all my intermediate thinking work anywhere: instead thinking afresh when a similar problem came up later helped in coming up with better solutions. I did occasionally save my thinking in design docs, but the equivalent to that is CLAUDE.md and similar human-reviewed markdown saved at explicit -umm- checkpoints.
Actually interesting, but how's that different from just putting your learning / decision context into the normal commit text (body) ? An LLM can search that too, and doesn't require a new cli tool.
EDIT: Or just keep a proper (technical) changelog.txt file in the repo. A lot of the "agentic/LLM engineering frameworks" boil down to best approaches and proper standards the industry should have been following decades ago.
After I have an ai dona task, I ask the next one to look at that plan and git diff and so ble check validate
I don't see the need for a full platform that is separate from where my code already lives. If I'm migrating away, it's to something like tangled, not another VC funded company
This is a good idea but I feel like you could get something similar by just adding an instruction for the agent to summarize the context for the commit into a .context/commit/<sha> file as a git hook.
Exactly. I don't want to wade through a whole session log just to get to reasoning, and more importantly, I don't want to taint my current agent context with a bunch of old context.
Context management is still an important human skill in working with an agent, and this makes it harder.
Feature plans generated by agents are often transient documents that fall away once the plan is executed. Ideally, that artifact would be preserved alongside the implementation.
My experience is that Cursor's reliance on VS Code's clunky panel-based UI and jack-of-all-trades harness is holding it back. Likewise, Claude Code shoe-horning a GUI into a TUI and perma-binding to a single model family is not the ideal end-state.
The VC play here? The git context CLI thing is a foundational step that lays the groundwork for a full IDE/workflow tool, I guess.
If you don't have a record of questions asked/answered and rationale for decisions taken, I've noticed it's easy for subsequent feature plans to clash. Maintaining a line of consistency across each feature plan is a good thing.
Is this the product? I don't want to jump on the detractor wagon, but I read the post and watched the video, and all I gathered is that it dumps the context into the commit. I already do this.
I guess if I had to do it, I'd reject pushes without the requisite commit to entire/checkpoints/v1. I think their storage strategy is a bit clunky for that, but it can be done. I might look to do something more like the way jujutsu colocates its metadata. I don't think this particular implementation detail makes too much of a difference, though. I got along just fine in a regulated environment by setting a policy and enforcing it before git existed. Ideally, we'd do things for a good reason, and if you can't get along in that world, then it's probably not the right job for you. Sometimes you've got to get the change controls in order before we can mainline your contributions because we have to sustain audits. I don't think this is about forcing people to do something that they otherwise wouldn't do if you told them that it's a requirement of the job.
100%. Day one is to ship the basic capability, which many of us have already vibe-coded... Day two is all the enterprise stuff to make big companies trust AI coding more. That could unlock a lot of revenue. This isn't random at all.
Git is totally fine keeping a few extra text files. These are ephemeral anyway. The working sessions just get squashed down and eliminated by the time I've got something worth saving anyway. At that point, I might keep a overview file around describing what the change does and how it was implemented.
(I will give the agent boom a bit of credit: I write a lot more documentation now, because it's essentially instruction and initial instruction to anything else that works on it. That's a total inversion, and I think it's good.)
The bigger problem is, like others have said, there's no one true flow. I use different agents for different things. I might summarize a lot of reasoning with a cheap model to create a design document, or use a higher reasoning model to sanity check a plan, whatever. It's a lot like programming in English. I don't want my tool to be prescriptive and imposing its technical restrictions on me.
All of that aside: it's impossible that this tool raised $60 million. The problem with this post is that it's supposed to be a hype post about changing the game "entirely" but it doesn't give us a glimpse into whatever we're supposed to by hyped about.
I have it (claude, codex) summarise what we've discussed about a design, big change, put it in an MD file and then I correct it, have it re-read it and then do the change.
Then later if it goes off piste in another session tell it to re-read the ADDs for x, y and z.
If someone could make that process less clunky, that would be great. However it's very much not just funnel every turd uttered in the prompt onto a git branch and trying a chug the lot down every session.
Very similar for me. I have a plans folder in my root where I store the plans while they're either under improvement or under implementation. Once they're done they're moved into the plans/old folder. So far it's worked great. It's a couple of manual steps extra but very helpful record.
Pretty much the same thing. I don't find it to be a burden. Regarding the product, I'm willing to believe I just don't see big picture, but without some peek at the magic, I don't know how much easier this could really be.
Huh, the checkpoint primitive is something that I've been thinking about for a while, excited to see how it's implemented in the CLI. Git-compatible structures seem to be a pretty big pull whenever they're talking about context management.
Already at a $300 million valuation at seed. Who’s going to join this as a regular IC who gets a pittance in options? I can’t imagine the risk matches the modest potential upside.
They'll raise at double that or more before the end of the year. The dynamics of the VC market right now are staggering to watch, but the money velocity is real, and this has "ex-CEO of Github" plus "AI".
It's legit mania in VC world even as they're looking at each going "is this mania? Is it mania if you're asking if it's mania". The only rule right now is the music is playing so no-one wants to grab a chair. There's a sense this might come crashing down, but what's a player gonna do, sit on the side while paper markups are producing IRR that is practically unprecedented?
I like the concept, but it seems like an obvious extension of where Anthropic is building with Tasks and Memory. What’s the moat? Cross-agent compatibility?
I would argue the largest developer tools/services acquisition in recent history was GitHub (assuming you don't count RedHat in that category). GitHub was primarily an acquisition focused on acquiring the network of developers, based on the trust and cachet GitHub had earned, with the intent to upsell (land and expand) the developers at large Enterprise companies on downstream workflows and tools.
If I were an employee looking to join Entire, or a developer evaluating the durability of Entire for my needs over the long-term, I'd ask things like —
- What's the terminal value of a DevTool in the AI era? Is it closer to a standard 10x ARR? or maybe 100x...perhaps 1000x?
- Is there value in the asset attributable to the network? If so, is it defensible? What is the likelihood protocols emerge that simply disintermediate the network moat of a AI agent memory company like Entire?
- What kind of AI data are developers willing to silo with a single vendor? What kind of controls do Enterprises expect of an AI agent memory system? Can Entire reasonably provide them to grow in the next 12-24 months?
- As a potential employee...if you join a company with a $60M seed funding and 19 employees, what is the ARR they need to achieve based on the first product in market in roughly ~12 months? $6M...$12M...20M? Is there a DevTools company that's ever done that!? What is the thesis for "this time is different"? Getting developers to pay for things is tough 'doncha know?
Only then can you ask basic technical diligence questions —
- Is Git a system that can scale to be the memory system and handle the kind of tasks that AI agents are expected handle?
- Are the infrastructure costs for these early platform decisions here reasonable over the long-term? Or do they just run out of money?
I wish them the best, and I hope employees get liquidity, and people take money off the table in a responsible way.
I don't understand how this is different from giving an agent access to github logs? The landing page is terrible at explaining what it does.I guess they are just storing context in git aswell?
So is this just a few context.md files that you tell the agent to update as you work and then push it when you are done???
I hacked together something similar to the concept they describe a few months ago (https://github.com/btucker/agentgit) and then ended up not actually finding it that useful and abandoning it.
I feel like the value would be in analyzing those rich traces with another agent to extract (failure) patterns and learnings, as part of a flywheel setup. As a human I would rarely if ever want to look at this -- I don't even have time to look at the final code itself!
> value would be in analyzing those rich traces with another agent to extract (failure) patterns and learnings
Claude Code supports hooks. This allows me to run an agent skill at the end of every agent execution to automatically determine if there were any lessons worth learning from the last session. If there were. new agent skills are automatically created or existing ones automatically updated as apporpriate.
Completely agree. But I wonder how much of that is just accomplished with well placed code comments that explain the why for future agent interactions to prevent them from misunderstanding. I have something like this in my AGENTS.md.
Checkpoints sounds like an interesting idea, and one I think we'll benefit from if they can make it useful.
I tried a similar(-ish) thing last year at https://github.com/imjasonh/cnotes (a Claude hook to write conversations to git notes) but ended up not getting much out of it. Making it integrated into the experience would have helped, I had a chrome extension to display it in the GitHub UI but even then just stopped using it eventually.
I had a similar, admitted poorly thought out idea a few months back.
I wanted to more or less build Jira for agents and track the context there.
If I had to guess 60 million is just enough to build the POC out. I don't see how this can compete though, Open AI or Anthro could easily spin up a competitor internally.
You don't need to build anything. Just tell the agent to write tickets into .md files in a folder and move them to a closed foler as you go along. I've been using Claude Code with the Max plan nonstop essentially every day since last July and since then I've come to realize that the newer people are the more things they think they need to add to CC to get it work well.
Eventually you'll find a way that works for you that needs none of it, and if any piece of all the extras IS ever helpful, Anthropic adds it themselves within a month or two.
I'm thinking a customized LLM would write notes in its own hyper compressed language which would allow it to be much much more efficient.
For debugging you could translate it out to English, but if these agents can do stuff without humans in the loop, why do they need to take notes in English?
I can't imagine creating this without hundreds of millions if not billions. I think the future is specialized models
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more.
Whether or not useful for agent collaboration, the data here will be more valuable than gold for doing RL training later on.
disclosure: i run a startup that will most likely be competitive in the future.
I welcome more innovation in the code forge space but if you’re looking for an oss alternative just for tracking agent sessions with your commits you should checkout agentblame
Did you have to choose an adjective to name your product. Now it’s going to be very confusing for search engines and LLms.
“Tell me more about entire.”
“Entire what?”
“You know, that entire thing.”
Another of your competitors here. It makes me giggle that we're going after the entire developer experience while Entire is only looking at a small corner of it.
Certainly! But just to confirm, you aren't making an IDE or building a version control system to replace Git, are you? While money means you need not fear me, the scale of my vision means that I don't fear you either.
I shall give the benefit of a doubt given they are "building in the open". I feel my current setup already does all this though, so I struggle to see the point
It’s funny. The whole “review intent", "learning" from past mistakes, etc, is exactly what my current set up does too. For free. Using .md files said agents generate as they go.
love the shout but git-ai is decidedly not trying to replace the SCMs. there are teams building code review tools (commercial and internal) on top of the standard and I don't think it'll be long before GitHub, GitLab and the usual suspects start supporting it since folks the community have already been hacking it into Chrome extensions - this one got play on HN last week https://news.ycombinator.com/item?id=46871473
This feels a bit like when some Hubbers broke off to work on PlanetScale, except without the massively successful, proven-to-be-scalable open source tool to build off (Vitess).
If you're approaching this problem-space from the ground up, there are just so many fundamental problems to solve that it seems to me that no amount of money or quality of team can increase your likelihood of arriving at enough right answers to ensure success. Pulling off something like this vision in the current red-ocean market would require dozens of brilliant ideas and hundreds of correct bets.
Just have a data lake with annotated agent sessions and tool blobs (you should already be keeping this stuff for evals), then give your agent the ability to query it. No need for a special platform, or SaaS.
As for SDLC, you can do some good automations if you're very opinionated, but people have diverse tastes in the way they want to work, so it becomes a market selection thing.
I'm interested to see if they will try to tackle the segregation of human vs AI code.
The downside of agents is that they make too much changes to review, I prefer being able to track which changes I wrote or validated from the code the AI wrote.
For people trying to understand the product (so far), it seems that entire is essentially an implementation of the idea documented by http://agent-trace.dev.
Sorry for not contributing to the discussion (as per the guidelines), but is it just me or this blog post reads a lot like LLM-filled mumble jumble? Seems like I could trim half of the words there and nothing would be lost.
New agent framework / platform every week now. It's crazy how fast things move...just when you get comfortable with an AI flow something new comes out...
I am already overloaded with information (generated by AI and humans) on my day to day job, why do I need this additional context, unless company I work for just wants to spend more money to store more slop?
How is it different than reversing it, given a PR -> generate prompt based on business context relevant to the repo or mentioned issues -> preserve it as part of PR description
I barely look at git commit history, why should I look for even higher cardinality data, in this case: WTF, are you doing, idiot, I said don't change the logic to make tests pass, I said properly write tests!
I love this. Prompt engineering is a real skill, and learning by example is the best way to do it. Juniors being able to see how seniors actually did something is really powerful.
I don't see how we need a brand new paradigm just because LLMs evidently suck at sharing context in their Git commits. The rules for good commits still apply in The New Age. Git is still good enough, LLMs (i.e. their developer handlers) just need to leverage it.
Personally, I don't let LLMs commit directly. I git add -p and write my own commit messages -- with additional context where required -- because at the end of the day, I'm responsible for the code. If something's unclear or lacks context, it's my fault, not the robot's.
But I would like to see a better GitHub, so maybe they will end up there.
With the how AI companies are advertising we can just tell the AI what we want and it will be done with no additional human interaction needed, why do we need a new type of development platform? We shouldn't need to collaborate at all.
Essentially all software is augmented with agentic development now, or if not, built with technology or on platforms that is
It's like complaining about the availability of the printing press because it proliferated tabloid production, while preferring beautifully hand-crafted tomes. It's reactively trendy to hate on it because of the vulgar production it enables and to elevate the artisanal extremes that escape its apparent influence
It's really not as integral as you make it sound. If I make one PR on a widely used open source tool with a small fix, is most software development augmented by me?
Outside of simply not being true, the sentiment of what you're saying isn't much different than:
"Essentially all software is augmented with Stack Overflow now, or if not, built with technology or on platforms that is."
Agentic development isn't a panacea nor as widespread as you claim. I'd wager that the vast majority of developers treat AI is a more specified search engine to point them in the direction they're looking for.
AI hallucination is still as massive problem. Can't tell you the number of times I've used agentic prompting with a top model that writes code for a package based on the wrong version number or flat out invents functionality that doesn't exist.
I just cannot fathom how people can say something like this today, agentic tools have now passed an inflection point. People want to point out the short comings and fully ignore that you can now make a fully functioning iPhone app in a day without knowing swift or front end development? That I can at my company do two projects simultaneously, both of them done in about 1/4 the time and one would not have even been attempted before due to the SWE headcount you would have to steal. There are countless examples I have in my own personal projects that just are such an obvious counter example to the moaning “I appreciate the craft” people or “yea this will never work because people still have to read the code” (today sure and this is now made more manageable by good quality agents, tomorrow no. No you won’t need to read code.)
I've found that the effort required to get a good outcome is roughly equal to the effort of doing it myself.
If I do it myself, I get the added bonus of actually understanding what the code is doing, which makes debugging any issues down the line way easier. It's also in generally better for teams b/c you can ask the 'owner' of a part of the codebase what their intuition is on an issue (trying to have AI fill in for this purpose has been underwhelming for me so far).
Trying to maintain a vibecoded codebase essentially involves spelunking though a non-familliar codebase every time manual action is needed to fix an issue (including reviewing/verifying the output of an AI tool's fix for the issue).
(For small/pinpointed things, it has been very good. e.g.: write a python script to comb through this CSV and print x details about it/turn this into a dashboard)
In sonnet 4 and even 4.5 I would have said you are absolutely right, and in many cases it slows you down especially when you don’t know enough to sniff trouble.
Opus 4.5 and 4.6 is where those instances have gone down, waaay down (though still true). Two personal projects I had abandoned after sonnet built a large pile of semi working cruft it couldn’t quite reason about, opus 4.6 does it in almost one shot.
You are right about learning but consider: you can educate yourself along the way — in some cases it’s no substitute for writing the code yourself, and in many cases you learn a ton more because it’s an excellent teacher and you can try out ideas to see which work best or get feedback on them. I feel I have learned a TON about the space though unlike when I code it myself I may not be extremely comfortable with the details. I would argue we are about 30% of the way to the point where it’s not even no longer relevant it’s a disservice to your company to be writing things yourself.
There are other things very good "at some range of common tasks". For example, stackoverflow snippets, libraries, bash spaghetti and even some no-code/low-code tools.
What part of Voyager I and Voyager II are "augmented with agentic development?"
Surely if all software is augmented with agentic development now, our most important space probes have had their software augmented too, right?
What about my blog that I serve static pages on? What about the xray machine my dentist uses? What about the firmware in my toaster? Does the New York Stock Exchange use AI to action stock trades? What about my telescope's ACSOM driver?
You’re talking about a 1970s satellite? I guess you win the argument?
Blog: I use AI to make and blog developers are using agentic tools
X-ray machine: again a little late here, plus if you want to start dragging in places that likely have a huge amount of beaurocracy I don’t know that that’s very fair
Firmware in your toaster: cmon these are old basic things, if it’s new firmware maybe? But probably not? These are not strong examples
NYSE to action on stock trades; no they don’t use AI to action on stock trades (that would be dumb and slow and horribly inefficient and non-deterministic), but may very well now be using AI to work on the codebase that does
Let’s try to find maybe more impactful examples than small embodied components in toasters and telescopes, 1970s era telescopes that are already past our solar system.
Im saying you’re missing the point and the spirit of the argument. Yes, you are right, voyager doesn’t use agentic AI! I don’t even think the other examples you used are as agentic free as you think. They may or may not be! What’s the point you want to make?
Hah. "If it's not too much trouble, would you mind if we disable the rimraf root feature?"
Gotta bully that thing man. There's probably room in the market for a local tool that strips the superfluous niceties from instructions. Probably gonna save a material amount of tokens in aggregate.
Which only reinforces someone just lit $60M on fire. It's trivial to do this and there are so many ways people do things, having the AI build custom for you is better than paying some VC funded platform to build something for the average
The lack of explanation of what it is and does is a tell of what gullible audience they are seeking.
Tech marketing has become a lot like dating, no technical explanation and intellectual honesty, just word words words and unreasonable expectations.
People usually cannot be honest in their romantic affairs, and here it is the same. Nobody can state: we just want to be between you and whatever you want to accomplish, rent seeking forever!
Will they ever care to elaborate HOW things works and the rationale behind stating this provides any benefit whatsoever? Perhaps this is not intended for those type of humans that care about understanding and logic?
$60M seed to wrap git hooks in YAML config. The AI tooling bubble is just VCs subsidizing solutions looking for problems while developers want less complexity, not more.
I really hate this trend of naming companies using dictionary words just because they can afford to spend cash on the domain name instead of engineering. Render, fly, modal, entire and so on.
I did test it and use it and trashed it because there is very little value, actually none for me. These problems are easily being solved in other ways whoever has any experience with these tools. Getting $60M round for this stuff is ridiculous.
The AI fatigue is real, and the cooling-off period is going to hurt. We’re deep into concept overload now. Every week it’s another tool (don’t get me started on Gas Town) confidently claiming to solve… something. “Faster development”, apparently.
Unless you’re already ideologically committed to this space, I don’t see how the average engineer has the energy or motivation to even understand these tools, never mind meaningfully compare them. That’s before you factor in that many of them actively remove the parts of engineering people enjoy, while piling on yet another layer of abstraction, configuration, and cognitive load.
I’m so tired of being told we’re in yet another “paradigm shift”. Tools like Codex can be useful in small doses, but the moment it turns into a sprawling ecosystem of prompts, agents, workflows, and magical thinking, it stops feeling like leverage and starts feeling like self-inflicted complexity.
> I don’t see how the average engineer has the energy or motivation to even understand these tools, never mind meaningfully compare them
This is why I use the copilot extension in VS code. They seem to just copy whatever useful thing climbs to the surface of the AI tool slop pile. Last week I loaded up and Opus 4.6 was there ready to use. Yesterday I found it has a new Claude tool built in which I used to do some refactoring... it worked fine. It's like having an AI tool curator.
Your point about the overwhelming proliferation of AI tools and not knowing which are worth any attention and which are trash is very true I feel that a lot today (my solution is basically to just lean into one or two and ask for recommendations on other tools with mixed success).
The “I’m so tired of being told we’re in another paradigm shift” comments are widely heard and upvoted on HN and are just so hard to comprehend today. They are not seeing the writing on the wall and following where the ball is going to be even in 6-12 months. We have scaling laws, multiple METR benchmarks, internal and external evals of a variety of flavors.
“Tools like codex can be useful in small doses” the best and most prestigious engineers I know inside and outside my company do not code virtually at all. I’m not one of them but I also do not code at all whatsoever. Agents are sufficiently powerful to justify and explain themselves and walk you through as much of the code as you want them to.
Yeah, I’m not disputing that AI-assisted engineering is a real shift. It obviously is.
My issue is that we’ve now got a million secondary “paradigm shifts” layered on top: agent frameworks, orchestration patterns, prompt DSLs, eval harnesses, routing, memory, tool calling, “autonomous” workflows… all presented like you’re behind if you’re not constantly replatforming your brain.
Even if the end-state is “engineers code less”, the near-term reality for most engineers is still: deliver software, support customers, handle incidents, and now also become competent evaluators of rapidly changing bot stacks. That cognitive tax is brutal.
So yes, follow where the ball is going. I am. I’m just not pretending the current proliferation is anything other than noisy and expensive to keep up with.
Really struggling to figure out what this is at a glance. Buried in the text is this line which I think is the tl;dr:
"As a result, every change can now be traced back not only to a diff, but to the reasoning that produced it."
This is a good idea, but I just don't see how you build an entire platform around this. This feels like a feature that should be added to GitHub. Something to see in the existing PR workflow. Why do I want to go to a separate developer platform to look at this information?
I'm sure i'm missing something but can you not ask the llm to add the reasoning behind the commit in the comments as part of the general llm instructions?
Oh man I'm tired. This reminds me of the docker era. It's all moving fast. Everyone's raising money. And 24 months from now it's all consolidating. It's all a nice hype game when you raise the funding but the execution depends on people finding value in your products and tools. I would argue yes many of these things are useful but I'd also argue there's far too much overlap, too many unknowns and too many people trying to reinvent the whole process. And just like the container era I think we're going to see a real race to zero. Where most of the dev tools get open sourced and only a handful of product companies survive, if that. I want to wish everyone the best of luck because I myself have raised money and spent countless years building Dev tools. This is no easy task especially as the landscape is changing. I just think when you raise $60m and announce a cli. You're already dead, you just don't know it. I'm sorry.
I see the value since I built a similar tool different approach. Then there's Beads, which is what inspired my project, with some tens of thousands of developers using it or more now? I'm not sure how they figure how many users they have.
In my case I don't want my tools to assume git, my tools should work whether I open SVN, TFS, Git, or a zip file. It should also sync back into my 'human' tooling, which is what I do currently. Still working on it, but its also free, just like Beads.
I wouldn't wanna be in the rat race myself, but I know people who salivate at the opportunity to create some popular dev tool to get acquired by MS, Google or Amazon or whichever of the big tech companies that decide this could work well in their cloud ecosystem.
On the one hand they think these things provide 1337x productivity gains, can be run autonomously, and will one day lead to "the first 1 person billion dollar company".
And in complete cognitive dissonance also somehow still have fantasies of future 'acquisition' by their oppressors.
Why acquire your trash dev tool?
They'll just have the agents copy it. Hell, you could even outright steal it, because apparently laundering any licensing issues through LLMs short circuits the brains of judges to protohuman clacking rocks together levels.
I think there are 2 parts here. That persona you’re describing (startup cofounder or engineer being paid mostly in equity) is a good subset of the people here. If I had to pull a number out of my shiny metal ass, I’d say it’s 30%. Those people both loath big tech, and dream of the day they are acquired by it. It’s not really the contradiction you think it’s. Another 45% of people here are tech-savvy Reddit refuges who say Reddit things.
As to why would those company acquire a startup instead of having an agent generate it for them. Why has big tech ever acquired tech startups when they could have always funded it in house? It’s not always a technical answer. Sometimes it’s internal Political fights, time to market, reduce competition, PR reasons or they just wanna hire the founder to lead a team for that internally and the only way he’ll agree is if there is an exit plan for his employees. I sat in “acquire or build” discussions before. The “how hard would it be to just do that?” Was just one of many inputs into the discussion. Ever wondered why big big companies acquire a smaller one, not invest in it, then shut it down few years later?
With 60 million you could have waited for a bigger announcement? There's "AI fatigue" among the target market for these sorts of tools, advertising unfinished products will take its toll on you later.
Chris was also CEO from 2008 to 2012. Tom had 2012 to 2014.
Nat's company Xamarin was acquired by Microsoft in 2016.
HockeyApp wasn't A/B testing, but a platform for iPhone, Mac, Android, and Windows Phone developers to distribute their beta version (like what TestFlight is today to the App Store), collect crash reports (like what Sentry is today), user feedback, and basic analytics for developers.
The Ximian thing I wrote from obviously faulty memory (I now wonder if it was influenced by early 2000s Miguel's bonobo obsession), the rest from various google searches. Should have gone deeper.
Ximian, Inc. (previously called Helix Code and originally named International Gnome Support) was an American company that developed, sold and supported application software for Linux and Unix based on the GNOME platform. It was founded by Miguel de Icaza and Nat Friedman in 1999 and was bought by Novell in 2003
...
Novell was in turn acquired by The Attachmate Group on 27 April 2011. In May 2011 The Attachmate Group laid off all its US staff working on Mono, which included De Icaza. He and Friedman then founded Xamarin on 16 May 2011, a new company to continue the development of Mono. On 24 February 2016, Microsoft announced that they had signed an agreement to acquire Xamarin.
Didn't say you were wrong. It was just missing the Xamarin step in the sequence of companies, and arguably Xamarin was the bigger milestone than Ximian.
We went from having new JavaScript frameworks every week to having new AI frameworks every week. I'm thinking I should build a HN clone that filters out all posts about AI topics...
Looking at the most popular agent skills, heavily geared towards react and JS, I think a lot of the most breathless reports of LLM success are weighted towards the same group of fashion-dependant JavaScript developers.
The same very online group endlessly hyping messy techs and frontend JS frameworks, oblivious to the Facebook and Google sized mechanics driving said frameworks, are now 100x-ing themselves with things like “specs” and “tests” and dreaming big about type systems and compilers we’ve had for decades.
I don’t wanna say this cycle is us watching Node jockies discover systems programming in slow motion through LLMs, but it feels like that sometimes.
Ironically, using LLM’s for React is an exercise in pain, because they’re all trained on the lowest common denominator. So even Opus is constantly fighting stupid reactivity bugs.
Just give me your bank account, claude API, Mother's maiden name, your zip code, your 3 digit security code, and anything else you think I might need to live as malfist the magnificant. Can I call you that?
I've long wished for a 'filter' feature for the hn feed -- namely the old trend of web3 slop -- but with little else than keywords to filter, it would likely be tedious and inaccurate. Ironically, I think with AI/LLMs it could be a little easier to analyze.
This is how software is being written now. What you propose is like joining a forum called "Small-Scale Manufacturing News" and filtering out all 3D-printing articles.
This sounds like a company idea someone just came up with yesterday off the cuff, pitched it, and got money for because of their credentials so no one can really say no to investing in it, despite nothing new or different? What's the service or product and how is it different than every 3rd Show HN?
The most interesting thing about everyone trying to position themselves as AI experts is the futility of it: the technology explicitly promises tomorrows models will be better then todays, which means the skill investment is deflationary: the best time to learn anything is tomorrow when a better model will be better at doing the same work - because you don't need to be (conversely if you're not good at debugging and reverse engineering now...)
the best time to learn anything is tomorrow when a better model will be better at doing the same work
doesn’t that presume no value is being delivered by current models?
I can understand applying this logic to building a startup that solves today’s ai shortcomings… but value delivered today is still valuable even if it becomes more effective tomorrow.
I think it also presumes that the skills of today won't be helpful in making you better, faster, stronger at knowing what to learn tomorrow. Skateboarding ain't snowboarding but I guarantee the experience helps.
I'm pretty much just rawdogging Claude Code and opencode and I haven't bothered setting up skills or MCPs except for one that talks to Jira and Confluence. I just don't see the point when I'm perfectly capable of writing a detailed prompt with all my expectations.
The problem is that so many of these things are AI instructing AI and my trust rating for vibe coded tools is zero. It's become a point of pride for the human to be taken out of the loop, and the one thing that isn't recorded is the transcript that produced the slop.
I mean, you have the creator of openclaw saying he doesn't read code at all, he just generates it. That is not software engineering or development, it's brogrammer trash.
I think the rationale is that with the right tools you can move much faster, and not burn everything to the ground, than just rawdogging Claude. If you haven't bothered setting up extra tools you may still be faster / better than old you, but not better than the you that could be. I'm not preaching, that's just the idea.
> That is not software engineering or development, it's brogrammer trash.
Yes, but it's working. I'm still reading the code and calling out specific issues to Claude, but it's less and less.
That’s true for “tips and tricks” knowledge like “which model is best today” or “tell the model you’ll get fired if the answer is wrong to increase accuracy” that pops up on Twitter/X. It’s fleeting, makes people feel like “experts”, and doesn’t age well.
On the other hand, deeply understanding how models work and where they fall short, how to set up, organize, and maintain context, and which tools and workflows support that tends to last much longer. When something like the “Ralph loop” blows up on social media (and dies just as fast), the interesting question is: what problem was it trying to solve, and how did it do it differently from alternatives? Thinking through those problems is like training a muscle, and that muscle stays useful even as the underlying technology evolves.
It does seem like things are moving very quickly even deeper than what you are saying. Less than a year ago langchain, model fine tuning and RAG were the cutting edge and the “thing to do”.
Now because of models improving, context sizes getting bigger, and commercial offerings improving I hardly hear about them.
> what problem was it trying to solve, and how did it do it differently from alternatives?
Sounds to me like accidental complexity. The essential problem is to write good code for the computer to do it's task?
There's an issue if you're (general you) more focused on fixing the tool than on the primary problem, especially when you don't know if the tool is even suitable,
You nailed it. Thats exactly how I feel. Wake me up when the dust settles, and i'll deep dive and learn all the ins and outs. The churn is just too exhausting.
I don't get the pressure. I don't know about you, but my job for a long time has been continually learning new systems. I don't get how so many of my peers fall into this head trip where they think they are gonna get left behind by what amounts to anticipated new features from some SaaS one day.
How do you both hold that the technology is so revolutionary because of its productive gains, but at the same time so esoteric that you better be ontop of everything all the time?
This stuff is all like a weird toy compared to other things I have taken the time to learn in my career, the sense of expertise people claim at all comes off to me like a guy who knows the Taco Bell secret menu, or the best set of coupons to use at Target. Its the opposite of intimidating!
I may just be a "doomer", but my current take is we have maybe 3-5 years of decent compensation left to "extract" from our profession. Being an AI "expert" will likely extend that range slightly, but at the cost of being one of the "traitors" that helps build your own replacement (but it will happen with or without you).
> the technology explicitly promises tomorrows models will be better then todays, which means the skill investment is deflationary
This is just wrong. A) It doesn’t promise improvement B) Even if it does improve, that doesn’t say anything about skill investment. Maybe its improvements amplify human skill just as they have so far.
I have a reading list of a bunch of papers i didn't get through over the past 2 years. it is crazy how many papers on this list are completely not talked about anymore.
I kinda regret going through the SeLU paper lol back in the late 2010s.
Either the business makes a profit before it gets swept away, or it doesn't. This should be your goal: make money before your business dies. If you do that, you succeeded. Businesses are always ephemeral.
This is very well put. I think this platform can be useful but I doubt it can be something as big as the think it will be. At the end of the day it’s just storing some info with your data. I guess they are trying to be the next GitHub (and clearly have the experience :)). I doubt that success can be replicated today with this idea, even with $60 mil to burn
You’ve described a technology, not a solution to a clearly articulated problem that customers actually have. The problem that you have described is vague, and it’s unclear that it’s actually a problem at all. Finally, you don’t provide a persuasive and concrete argument about how your eventual solution—whatever that may be—will solve it.
I don’t mean to be so presumptuous as to teach Grampa how to suck eggs, but I think Amazon’s working backwards process is instructive.
Hey, is JJ compatibility in the cards? Considering the blog article hints at a goal of a developerless agent-to-agent automation platform I'm guessing developer conveniences are a side quest rn?
The founder has only forked repositories on GitHub that are sort of light web development related.
His use of bombastic language in this announcement suggests that he has never personally worked on serious software. The deterioration of GitHub under his tenure is not confidence inspiring either, but that of course may have been dictated by Nadella.
If you are very generous, this is just another GitHub competitor dressed up in AI B.S. in order to get funding.
Personal attacks aren't allowed here, so I've banned the account. If you want to use HN as intended, you'd be welcome to review https://news.ycombinator.com/newsguidelines.html and then email hn@ycombinator.com with a reason to believe that you'll follow the rules in the future.
Founder here. I built commercial insurance software for Windows 95 in the 1990s, driver assistant systems at Mercedes and at Bosch in the early 2000s, dozens of iPhone apps as contractor, a startup called HockeyApp (acquired by Microsoft), and various smaller projects, mostly in Ruby on Rails. And of course, when I left Microsoft & GitHub, 10 years of green boxes were removed from my GitHub profile.
Glad to see you commenting here despite the, er, unsubstantive disposition of some of the commenters! (I'm a mod here btw.) We really don't like it when the audience responds to new things with snark and dismissiveness, and there are rules in https://news.ycombinator.com/newsguidelines.html to make that clear.
In this case I think the root problem is that the OP (https://entire.io/blog/hello-entire-world/) is the wrong genre for HN. It's a fine fundraising announcement, but that sort of enthusiastic general announcement rubs the HN audience the wrong way because what they really want is technical details. Spectacular non technical details like high valuations, etc., tend to accentuate this gap.
I mention this because if you or someone on your team wants to write a technical post about what you're building, with satisfying details and all that, then we could do a take 2 (whenever would be a good time for this).
Makes sense, and we definitely will have more technical posts coming soon and on a regular basis. And I am German, I am used to snark. :P Joking aside, thanks for the response and appreciate your work as a mod!
It's sad to see that ex-GitHub CEO didn't make enough money to just kick-start his company, but needs external money which will later on dictate how the company works or will sell users and the product for the next exit..
So.. yea. Ignore and move on.
Every 'platform for AI agents' announcement makes me wonder: are we building tools for a workflow that actually exists, or are we building tools and hoping the workflow materializes? The GitHub parallel is instructive because GitHub succeeded by meeting developers where they already were (git). The question for Entire is whether agents have a natural coordination layer yet or if this is premature infrastructure.
I don't think your Github example is accurate. The vast majority of developers started using git after Github became a thing. They may have used svn or another type of collaboration system before, but not git. And the main reason they started using git is because Github was such massive value on top of git, not because git was so amazing.
I do not think that's how it worked out for GitHub: I'd rather say that Git (as complex as it was to use) succeeded due to becoming the basis of GitHub (with simple, clean interface).
At the time, there were multiple code hosting platforms like Sourceforge, FSF Savannah, Canonical's Launchpad.net, and most development was still done in SVN, with Git, Bazaar, Mercurial the upstart "distributed" VCSes with similar penetration.
Yes, development was being done in SVN but it was a huge pain. Continuous communication was required with the server (history lookups took ages, changing a file required a checkout, etc.) and that was just horribly inefficient for distributed teams. Even within Europe, much more so when cross-continent.
A DVCS was definitely required. And I would say git won out due to Linus inventing and then backing it, not because of a platform that would serve it.
Yes to all that. And GitLab the company was only founded in 2014 (OSS project started in 2011) and ran through YC in 2015, seven years after GitHub launched.
and most of those, except maybe gitlab, were clunky AF to use
Of the thousands, a handful will prevail. Most of it is vaporware, just like in any boom. Every single industry has this problem; copy-cats, fakes & frauds.
"Buy my fancy oil for your coal shovel and the coal will turn into gold. If you pay for premium, you don't have to shovel yourself."
If everything goes right, there won't be a coal mine needed.
I'd bet that less people had their source code on git in 2008 than the number of developers using the various coding agents today. And the open-source project that we published today hooks into the existing workflow for those developers, in Claude Code and in Gemini CLI. Time will tell the rest. We will publish regular updates and you can judge us on those results.
At least for me, I have felt like the chat history in an agent is often times just as important and potentially even more important than the source code it generates. The code is merely the compiled result of my explanations of intent and goals. That is, the business logic and domain expertise is trapped in my brain, which isn't very scalable.
Versioning and tracking the true source code, my thoughts, or even the thoughts of other agents and their findings, seems like a logical next step. A hosted central place for it and the infrastructure required to store the immense data created by constantly churning agents that arrive at a certain result seems like the challenge many seem to be missing here.
I wish you the best of luck with your startup.
The hype is the product
We are building tools and hoping an exit materializes. There’s so much funny money in AI right now, getting life-altering money seems easily attainable
HN is full of AI agents hype posts. I am yet to see legitimate and functional agent orchestration solving real problems, whether it is scale or velocity.
> Entire, backed by a $60 million
This is the point of that post and helpfully it was added at the top in a TL;dr and was half of that t sentence TL;dr. Will succeed or not? Well, that's a coin toss, always been.
I mean, pretty much all big startups begin as "niche" things that people might care about later. Tesla, Airbnb, Twitch... and countless failures too. It's just how the game is.
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more.
This thread is extremely negative - if you can't see the value in this, I don't know what to tell you.
I currently develop small utilities with the help of AI, but am far from vibe coding or using agents. I review every single suggestion and do some refactoring at each step, before any commit (sometimes heavy refactoring; sometimes reorganizing everything).
In my experience LLMs tend to touch everything all of the time and don't naturally think about simplification, centralization and separation of concerns. They don't care about structure, they're all over the place. One needs to breathe on their shoulders to produce anything organized.
Maybe there's a way to give them more autonomy by writing the whole program in pseudo-code with just function signatures and let them flesh it out. I haven't tried that yet but it may be interesting.
What kind of barrier/moat/network effects/etc would prevent someone with a Claude Code subscription from replicating whatever "innovation" is so uniquely valuable here?
It's somewhat strange to regularly read HN threads confidently asserting that the cost of software is trending towards zero and software engineering as a profession is dead, but also that an AI dev tool that basically hooks onto Git/Claude Code/terminal session history is worth multiples of $60+ million dollars.
I built a basic copy with about an hour with my own "platform for ai agents" that I built out over the last week: https://github.com/jwbron/egg/pull/504, and refined it here: https://github.com/jwbron/egg/pull/517 (though right after I merged this I blew through my weekly token quota for my second claude max 20x account so I haven't been able to test it out yet).
I think your point is valid and I've been having the same thoughts. My tooling is still in the experimental phase, but I can move so quickly that I'm having trouble grasping how products like this will survive. If I can build this out in a week and copy an idea like this one (which is a great one, mind you) in an hour, what's the value of paying someone for a product like this vs just building it myself?
There's no way this company is just a few git and claude hooks with a CLI. They're definitely working on a SASS - something else that isn't open source that this primitive is the basis of. Like a GitHub for agent code
> What kind of barrier/moat/network effects/etc would prevent someone with a Claude Code subscription from replicating whatever "innovation" is so uniquely valuable here?
You are correct, that isn't the moat. Writing the software is the easy part
If they had wanted a moat for this part of their offering, they wouldn’t have open-sourced it.
This is not their offering, this is a tool to raise interest.
This comment feels word-for-word the legendary DropBox critique on HN.
The same moat that git had on svn, a better mental paradigm over the same fundamental system, more suited to how SWE changed over a decade.
git didn't succeed based on the mental model. It got a foot in the door with better tooling and developer experience then blew the door open when GitHub found a way to productize it.
Git doesn't have a moat. Git isn't commercial software, and doesn't need to strong arm you into accepting bad license terms.
> HN threads confidently asserting
I have never seen any thread that unanimously asserts this. Even if they do, having HN/reddit asserting something as evidence is wrong way to look at things.
HN is full of AI agents hype posts. I am yet to see legitimate and functional agent orchestration solving real problems, whether it is scale or velocity.
HN is full of anti hype posts as well. If I were to estimate there are more posts of anti hype than of hype.
For the last three or four months, what I've been doing is anytime I have Claude write a comment on an issue, it just adds a session ID, file path and the VM it is on. That way, whenever we have some stuff that comes up, we just search through issues and then we can also retrace the session that produced the work and it's all traceable. In general, I just work through gitea issues and sometimes beads. I couldn't stand having all these MD files in my repo because I was just drowning in documentation, so having it in issues has been working really nicely and agents know how to work with issues. I did have it write a gitea utility and they are pretty happy using/abusing it. Anytime I see that they call it in some way that generates errors, I just have them improve the utility. And by this point, it pretty much always works. It's been really nice.
Sure... you `git add` the context text generated by AI and `git commit` it, could be useful. Is that worth 60 million?
It’s good to know that a few decades later the same generic Dropbox-weekend take can be made.
Survivorship bias. How many failed and commenters were right?
predicting that a startup will fail is.. well, you got a ton of probability on your side there. so it isn't a particularly impressive thing to be right about.
Unimpressive doesn't mean incorrect, sometimes it's good to take the side of the most probable. And yet at the same time I am reminded of this quote:
> The reasonable man adapts himself to the world: the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. - George Bernard Shaw
Sometimes adapting oneself is, in fact, progress.
I'm not disagreeing, just soliciting. Does anyone have examples of products that failed in the early stages because their implementation was too trivial?
How exactly are we supposed to hear about something that failed in the early stages?
By listening to your friends and circle
99% of projects the take applies to are massive flops. The Dropbox weekend take is almost always correct.
Yeah I guess why would anyone build anything, 99% of projects are flops.
I mean a dev tool that's seemingly failing to resonate with developers as to why they would pay for this is a pretty good way to tell if it's going to fall in the 1%.
The Dropbox take was wrong because they didn't understand the market for the product. This time the people commenting are the target audience. You even get the secondary way this product will lose even if turns out to be a good idea, existing git forges won't want to lose users and so will standardize and support attaching metadata to commits.
People keep saying that, but it's hardly the same thing. We're talking about developer workflow here. It's like someone coming up with Brancher. It's a git branch manager. Use `brancher foo` to replace `git checkout -b foo`. "Remember that comment about rsync and dropbox? Brancher is to git, what dropbox is to rsync"
How is LangChain doing? How about OpenAI's Swarm or their Agent SDK or whatever they called it? AWS' agent-orchestrator? The crap ton of Agent Frameworks that came out 8-12 months ago? Anyone using any of these things today? Some poor souls built stuff on it, and the smart ones moved away, and some are stuck figuring out how to do complex sub-agent orchestration and handoffs when all you need apparently is a bunch of markdown files.
Just saw a Discord-weekend take on reddit! Haha. Guy was saying he could create it in a day and then self-host it on his servers so that he doesn't have to put Nitro ads on top of it
It's funny how HN'ers frequently judge ideas based on complexity of implementation, not value.
I still remember the reaction when Dropbox was created: "It's just file sharing; I can build my own with FTP. What value could it possibly create".
Dropbox value was instantly recognizable, but I feel I have zero use for Entire.
Discord is not prized because you can send a message to a chatroom, or any of the hooks and functions.
It's because of everybody there.
Currently no one is on Entire - the investor are betting they will be.
They raised 60 million. The investors think it’s worth 600M+
It's the valuation that is wild to me (I think the idea itself has merit). But these are the new economics. I can only say "that's wild" enough before it is in fact no longer wild.
These aren't new economics, it's just VC funds trying to boost their holdings by saying it's worth X because they said so. Frankly the FTC should make it illegal.
That's not how it works at all. Why stop at $300M, why didn't they just say $1BN out the gate?
That is where I’m shocked being in a position of raising for a startup myself, what was in their pitch deck/data room that convinced them of this valuation? Or is it due to the founders reputation and not the substance?
That's not impressive. That's an incredible amount concentrated in the hands of a few looking for a place to live. It has to end up somewhere. Some of it goes everywhere.
We have had this for ages now.... I just don't have access to the sort of people willing to pass me 60m for that. I never thought it to be worth anything really ; it was a trivial to implement afterthought.
I love this one so much! The arbitrary decision to cherry-pick critique a particular product to this degree, when it’s something that could be said about 99% of the stuff SV churns out, including in all likelihood anything you’ve ever worked on.
Good thing the comment you're replying to does not lionise 99% of the stuff SV churns out, including in all likelihood anything they've ever worked on. I guess we should just not critique anything out of SV because it's all shit?
Couldn't we capture this value with a git hook?
The unannounced web collaboration platform in-progress might be.
300 million, apparently.
That is their first feature.
If it were also their last, I would be inclined to agree.
Wow, read through the comments and you weren't joking. I attribute this to crossroads of "this release is v0.1 of what we are building" and the HN crowd who have been scrolling past 120 AI frameworks and hot takes daily and have no patience for anything that isn't immediately 100% useful to them in the moment.
I find the framing of the problem to be very accurate, which is very encouraging. People saying "I can roll my own in a weekend" might be right, but they don't have $60M in the bank, which makes all the difference.
My take is this product is getting released right now because they need the data to build on. The raw data is the thing, then they can crunch numbers and build some analysis to produce dynamic context, possibly using shared patterns across repos.
Despite what HN thinks, $60M doesn't just fall in your lap without a clear plan. The moat is the trust people will have to upload their data, not the code that runs it. I expect to see some interesting things from this in the coming months.
I haven't read the article yet but this conversation reminds me of Docker. Lots of people "didn't get it." I told them at the time: if you don't get it you aren't ready for it yet so don't worry about it. When you do need it, you'll get it and then you'll use it and never look back. Look at where we are with containers now.
And look where Docker Inc is now (which is one of the points some critics are making)
I know about "the entire developer world has been refactored" and all, but what exactly does this thing do?
Runs git checkpoint every time an agent makes changes?
100% agree because there’s a lot of value in understanding how and why past code was written. It can be used to make better decisions faster around code to write in the future.
E.g., if you’ve ever wondered why code was written in a particular way X instead of Y then you’ll have the context to understand whether X is still relevant or if Y can be adopted.
E.g., easier to prompt AI to write the next commit when it knows all the context behind the current/previous commit’s development process.
But that's not what is in the whole context. The whole context contains a lot of noise and false "thoughts". What the AI needs to do is to document the software project in an efficient manner without duplication. That's not what this tool is doing. I question the value in storing all the crap in git.
I wonder. How often will that context actually be that valuable vs just more bloat to fill up future API calls with to burn tokens.
A year ago I added memory to my Emacs helper [0]. It was just lines in org-mode. I thought it was so stupid. It worked though. Sort of.
That's how a trillion dollar company also does it, turns out.
0: https://github.com/karthink/gptel
ehhhh is it really that useful though? Sounds way more noisy than anything, and a great way to burn through tokens. It's like founding a startup to solve the problem of people squashing their commits. Also, it sounds like something Claude Code/Codex/etc could quickly add an extension for.
How would this use any extra tokens? Just seems like it's serializing the existing context
Maybe use critical thinking instead of a mindless dismissal?
The fact that you aren't haven't offered a single counterargument to any other posters' points and have to resort to pearl-clutching is pretty good proof that you can't actually respond to any points and are just emotionally lashing out.
This is literally what claude code already does minus the commit attachment. It’s just very fancy marketing speak for the exact same thing.
I’m happy to believe maybe they’ll make something useful with 60M (quite a lot for a seed round though), but Maybe not get all lyrical about what they have now.
Claude Code captures this locally, not in version control alongside commits.
I wonder how difficult it would be for Claude Code to have such a feature in a future release.
I see the utility in this as an extension to git / source control. But how do VCs make money of it?
I built out the same thing in my own custom software forge. Every single part of the collaborative development process is memoized.
And how are you using it now? Have you seen real value weeks or months on?
[flagged]
Please don't use quotation marks to make it look like you're quoting someone when you aren't. That's an internet snark trope and we're trying to avoid those on HN.
https://news.ycombinator.com/newsguidelines.html
Look it’s obvious at this point to anyone who is actually using the tools.
We can articulate it but why should we bother when it’s so obvious.
We are at an inflection point where discussion about this, even on HN, is useless until the people in the conversation are on a similar level again. Until then we have a very large gap in a bimodal distribution, and it’s fruitless to talk to the other population.
Not really, because those details aren't actually relevant to code archaeology.
You could have someone collect and analyze a bunch of them, to look for patterns and try to improve your shared .md files, but that's about it
Some Tom Dick and Harry to VCs: I have a proposal for you.
VCs: what is it
Tom Dick & Harry: AI
VCs: get the ** out of here, we already burnt enough money and will never see it back
Tom Dick & Harry: hear me out this is different
VCs: ok you have 5 minutes to explain me your product
Tom Dick & Harry: I dont have one
VCs: get the ** out of here
Tom Dick & Harry: hear me out
VCs: ok, you have 30 seconds to impress us.
Tom Dick & Harry: I just quit Microslop and still have high level contacts there
VCs: Hot damn!!! you are our lottery ticket to recoup all the money we have lost in other ventures. This is going to be a race against time, before your contacts go stale. Here's 60M for you, wine and dine your friends with it. On your way out you will find some AI generated product names and some vague product descriptions. Pick one and slap it on some website and announce our deal. Now get the ** out of here.
The thread is missing the forest for the trees. The interesting bet here isn't git checkpoints—it's that someone is finally building the observability layer for agent-generated code.
Most agent frameworks (LangChain, Swarm, etc.) obsessed over orchestration. But the actual pain point isn't "how do I chain prompts"—it's "what did the agent do, why, and how do I audit/reproduce it?"
The markdown-files-in-git crowd is right that simple approaches work. But they work at small scale. Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs.
The $60M question is whether that problem is big enough to justify a platform vs. teams bolting on their own logging. I'm skeptical—but the underlying insight (agent observability > agent orchestration) seems directionally correct.
@dang with the launch of open claw I have seen so much more LLM slop comments. I know meta comments like mine aren't usually encouraged, but I think we need to do something about this as a community. Is there anything we can do? (either ban or at least requiring full disclosure for bot comments would be nice).
EDIT: I suspect the current "solution" is to just downvote (which I do!), but I think people who don't chat with LLMs daily might not recognize their telltale signs so I often see them highly upvoted.
Maybe that means people want LLM comments here, but it severely changes the tone and vibe of this site and I would like to at least have the community make that choice consciously rather than just slowly slide into the slop era.
The moltbook-ification of every online forum seems inevitable this year. I wish we had a counter to this.
Parent comment has the rhythm of an AI comment. Caught myself not realizing it until you mentioned it. Seems like I am more in tune with LLM slop on twitter, which is usually much worse. But on second sight it's clear and it also shows the comment as having no stance, and very generic.
@dang I would welcome a small secondary button that one can vote on to community-driven mark a comment as AI, just so we know.
It's the dead internet theory in action. Every time I see slop I comment on it. I've found people don't always like it when you comment on it.
Yes I usually just bite my tongue and downvote, but with the launch of open claw I think the amount of slop has increased dramatically and I think we need to deal with it sooner than later.
Do you really think openclaw is to blame? I shudder to think of how few protections HN has against bots like that.
Nothing about the parent comment suggests AI, except the em dash, but that's just a regular old punctuation that predates AI.
How much experience do you have interacting with LLM generated prose? The comment I replied to sets off so many red flags that I would be willing to stake a lot on it being completely LLM generated.
It's not just the em dashes - its the cadence, tone and structure of the whole comment.
Yeah it's really frustrating how often I see kneejerk rebuttals assuming others are solely basing it on presence of em-dashes. That's usually a secondary data point. The obvious tells are more often structure/cadence as you say and by far most importantly: a clear pattern of repeated similar "AI smell" comments in their history that make it 100% obvious.
I didn’t catch it until seeing these flag-raising comments… checking the other comments from the last 8 hours, it’s Claw for sure.
Punchy sentence. Punchy sentence. It's not A, it's B.
The actual insight isn't C, it's D.
You're absolutely right! It's not the tooling, it's the platform.
This sounds awfully like an LLM generated comment.
I suppose it was just a matter of time before this kind of slop started taking over HN.
> Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs.
This has been the story for every trend empowering developers since year dot. Look back and you can find exactly the same said about CD, public cloud, containers, the works. The 'orchestration' (read compliance) layers always get routed around. Always.
I thought everyone was just using open telemetry traces for this? This is just a classic observability problem that isn’t unique with agents. More important yes, but not unique functionally.
Can you explain more how otel traces solve this problem? I don't understand how it's related.
Ok, I’ll grant you that if they can get agents to somehow connect to other’s reasoning in realtime that would be useful. Right now it’s me that has to play reasoning container.
This is interesting. I’m experimenting with something adjacent in an open source plugin, but focused less on orchestration and more on decision quality.
Instead of just wiring agents together, I require stake and structured review around outputs. The idea is simple: coordination without cost trends toward noise.
Curious how entire.io thinks about incentives and failure modes as systems scale.
That is a sharp observation———it is the observability that matters! The question arises: Who observes the observers? Would you like me to create MetaEntire.ai———an agentic platform that observes Entire.io?
I think you need a few more em-dashes there to be safe
I think we need an Agent EE Server Platform. :P
It's not this, it's that?
verbatim llm output with little substance to it. HN mods don't want us to be negative but if this is what we have to take serious these days it is hard to say anything else.
I guess I could not comment at all but that feels like just letting the platform sink into the slopacolypse?
A. B isn't C—it's D1.
E. But F, G: H1, H2...
I. J—but D2 seems K.
Yes—it is!
Wholeheartedly agree. We have been working hard at a solution towards this and welcome any feedback and skepticism: https://github.com/backbay-labs/clawdstrike
I have an agent write a file with this template each run:
```markdown # Run NNNN
## First Impressions [What state is the project in? What did the last agent leave?]
## Plan [What will you work on this iteration? Why?]
## Work Log [Fill this in as you work]
## Discoveries [What did you learn? What surprised you? What should the next agent know?]
## Summary [Fill this in before committing] ```
This is surprisingly effective and lets agents easily continue in progress work and understand past decisions.
I have CURRENT_TASK.md that does more or less the same thing. It also gets committed to git. So I guess that’s entire? Wish I’d realized I was sitting on a 60M idea…
> Checkpoints run as a Git-aware CLI. On every commit generated by an agent, it writes a structured checkpoint object and associates it with the commit SHA. The code stays exactly the same, we just add context as first-class metadata. When you push your commit, Checkpoints also pushes this metadata to a separate branch (entire/checkpoints/v1), giving you a complete, append-only audit log inside your repository. As a result, every change can now be traced back not only to a diff, but to the reasoning that produced it.
The context for every single turn could in theory be nearly 1MB. Since this context is being stored in the repo and constantly changing, after a thousand turns, won't it make just doing a "git checkout" start to be really heavy?
For example, codex-cli stores every single context for a given session in a jsonl file (in .codex). I've easily got that file to hit 4 GB in size, just working for a few days; amusingly, codex-cli would then take many GB of RAM at startup. I ended up writing a script that trims the jsonl history automatically periodically. The latest codex-cli has an optional sqlite store for context state.
My guess is that by "context", Checkpoints doesn't actually mean the contents of the context window, but just distilled reasoning traces, which are more manageable... but still can be pretty large.
To add to your comment, I think the next logical question is, “then what?” Surely one can’t build a sustainable business storing these records alone.
MCP server with RAG to feed it back to agents when they are working on a piece of code, and bob's your uncle
> won't it make just doing a "git checkout" start to be really heavy?
not really? doesn't git checkout only retrieve the current branch? the checkpoint data is in another branch.
we can presume that the tooling for this doesn't expect you to manage the checkpoint branch directly. each checkpoint object is associated with a commit sha (in your working branch, master or whatever). the tooling presumably would just make sure you have the checkpoints for the nearby (in history) commit sha's, and system prompt for the agent will help it do its thing.
i mean all that is trivial. not worth a $60MM investment.
i suspect what is really going on is that the context makes it back to the origin server. this allows _cloud_ agents, independent of your local claude session, to pick up the context. or for developer-to-developer handoff with full context. or to pick up context from a feature branch (as you switch across branches rapidly) later, easily. yes? you'll have to excuse me, i'm not well informed on how LLM coding agents actually work in that way (where the context is kept, how easy it is to pick it back up again). this is just a bit of opining based on why this is worth 20% of $300MM.
if i look at https://chunkhound.github.io it makes me think entire is a version of that. they'll add an MCP server and you won't have to think about it.
finally, because there is a commit sha association for each checkpoint, i would be worried that history rewrites or force pushes MUST use the tooling otherwise you'd end up screwing up the historical context badly.
This feels interesting because the real problem with AI generated code isn't that it is of better or worse quality than code written by a human, it's that we humans need to audit both types of code. And this platform feels like it addresses this in a novel and traditional way. I like it.
> Spec-driven development is becoming the primary driver of code generation.
This sounds like my current "phase" of AI coding. I have had so many project ideas for years that I can just spec out, everything I've thought about, all the little ideas and details, things I only had time to think about, never implement. I then feed it to Claude, and watch it meet my every specification, I can then test it, note any bugs, recompile and re-test. I can review the code, as you would a Junior you're mentoring, and have it rewrite it in a specific pattern.
Funnily enough, I love Beads, but did not like that it uses git hooks for the DB, and I can't tie tickets back to ticketing systems, so I've been building my own alternative, mine just syncs to and from github issues. I think this is probably overkill for whats been a solved thing: ticketing systems.
I am going lower level - every individual work item is a "task.md" file, starts initially as a user ask, then add planning, and then the agent checks gates "[ ]" on each subtask as it works through it. In the end the task files remain part of the project, documenting work done. I also keep an up to date mind map for the whole project to speed up start time.
And I use git hooks on the tool event to print the current open gate (subtask) from task.md so the agent never deviates from the plan, this is important if you use yolo mode. It might be an original technique I never heard anyone using it. A stickie note in the tool response, printed by a hook, that highlights the current task and where is the current task.md located. I have seen stretches of 10 or 15 minutes of good work done this way with no user intervention. Like a "Markdown Turing Machine".
If you are using claude, in your project's `.claude/settings.json` you can add something like:
```
```Just update it to iterate over your file. It should be a little easier to manage than git hooks and can hammer in testing.
That's hilarious, I called it gates too for my reimplementation of Beads. Still working on it a bit, but this is the one I built out a month back, got it into git a week ago.
For me a gate is: a dependency that must pass before a task is closed. It could be human verification, unit testing, or even "can I curl this?" "can I build this?" and gates can be re-used, but every task MUST have one gate.
My issue with git hooks integration at that level is and I know this sounds crazy, but not everyone is using git. I run into legacy projects, or maybe its still greenfield as heck, and all you have is a POC zip file your manager emailed you for whatever awful reason. I like my tooling to be agnostic to models and external tooling so it can easily integrate everywhere.
Yours sounds pretty awesome for what its worth, just not for me, wish you the best of luck.
https://github.com/Giancarlos/GuardRails
that's similar to the workflow i built, inspired by Recursive Language Models: https://github.com/doubleuuser/rlm-workflow
This is built in to Claude Code, when you're in plan mode it makes a task MD file, even giving it a random name and storing it in your .claude folder.
I'm confused how this is any different to the pretty standard agentic coding workflow?
Me too. I've been using spec-kitty [0], a fork of Spec Kit. Quite amazing how a short interview on an idea can produce full documents of requirements, specs, tasks, etc. After a few AI projects, this is my first time using spec driven development, and it is definitely an improvement.
[0]: https://github.com/Priivacy-ai/spec-kitty
Nice, I'll check yours out after work, looks pretty polished.
Task management is fundamentally straightforward and yet workflow specific enough that I recommend everyone just spend a few hours building their own tools at this point.
Beads is a nightmare.
I started off with the original beads and it was definitely a nightmare. However I would recommend using https://github.com/Dicklesworthstone/beads_rust - it's a much simpler implementation of the same concept, without all the random extra stuff thrown on to support Gas Town.
Ignoring the VC economics and awful name, I won’t be as pessimistic as everyone. I see the vision.
That said, nobody knows what the AI future looks like. Entire’s entire thesis is a solution for something we don’t even know we need. It’s a massive bet and uphill battle. Traditionally, dev tool success stories come from grassroots projects of developers solving their own problems and not massive VC funded efforts that tell you what you need to do.
I feel like it says a lot if this is the not pessimistic take.
My pessimism is mostly rooted in the VC economics of it all. The vision is great, but its a busy space and there's no actual product or business. They basically wrote the guy a check to build the spaceship in space.
And yet, this is precisely what seed bets are about. You have to try it in order to know whether there is a "there" there.
$1.5M seed bets, maybe. not $60M though
> The game has changed. The system is cracking.
Just say what your thing does. Or, better yet, show it to me in under 60 seconds.
Web sites are the new banner ads and headings like that are the new `<blink>`.
Exactly ... tired by all the marketing hyperbole talk. Just show what your product does in a simple example / showcase. If it's good, people will like it. You can save yourself a lot of text copy and user time that way.
They'll learn soon enough that selling to developers necessitates speaking clearly.
Dohmke never spoke clearly to developers when he was GitHub's CEO.
A CEO is never speaking to developers, he's speaking to other CEOs.
CEOs have many audiences; great CEOs communicate capably with each.
FWIW it's not entirely clear to me who Entire's long-term customer is, but the (interesting!) CLI that shipped today is very much for developers who are busy building with agents.
They will sell to their managers
No. With this kind of bullshit they plan to try to sell to C-levels and board members.
Edit: Actually it may just be aimed at investors. Who cares about having a product?
> Actually it may just be aimed at investors
The fact that the first image you see has "$60M seed" in big text, I have to agree, this does not feel aimed at devs.
The problem is that when it comes to (commercial) developer tools and services, everyone can/wants to be everything, so why let a simple statement or a showcase limit you? "Hey, we are a container scanning service... But we can also be a container registry too, a CI, a KeyValue store, an agent sandbox provider, git hosting? We can do quick dev deployments/preview too. Want a private npm registry? Automated pull request reviews? Code Signing service? We are working on a new text editor btw"
But what if my product is just an attempt to make a cushy exit during the AI bubble?
I feel like these types of pages are less geared towards actual users of the product and more towards the investors who love the vague and flowery language. We're no longer in a world where the path to profitability was the objective goal anyway, it makes sense to me that the marketing of software is becoming decreasingly detached from reality..
It's almost like an extension of the "if you're not paying for the product, you are the product" idea. If you're assessing a tool like this and the marketing isn't even trying to communicate to you, the user, what the product does, aren't you also kind of "the product" in this case too?
> Web sites are the new banner ads and headings like that are the new `<blink>`.
It's been like this since the Dotcom era
Or did you forget that you can do anything at zombo.com?
It appears to be rather slow today, but here's a Wiki link for the uninitiated- https://en.wikipedia.org/wiki/Zombo.com
The domain expired a few days ago and was purchased by someone else and then changed. There's a recreation of the original here https://html5zombo.com/
That's the saddest news I've heard this year.
It's still around, but has been redesigned and it's under "new management". Further proof that the internet is dying.
Wait really?!? I’m surprised at how much that saddens me. What is the point of the internet without zombo com
I couldn’t figure out what they were doing in the first few screens of scrolling. Moved on.
Its like a modern day redux of zombo com.
That’s a bit insulting to zombo.com.
AI is everything at zombo.com.
Everything is AI at zombo.com.
They also seem bothered by color photography in 2026. All style, no substance.
You need to use AI to summarize the point of articles about AI products
Seems they install a Git hook or something that executes on commit and saves your chatbot logs associated with the commit hash. This is expected to somehow improve on the issue that people are synthesising much more code than they could read and understand, and make it easier to pass along a bigger context next time you query your chatbots, supposedly to stop them from repeating "mistakes" that have already wasted your time.
What it does? Imagine a multi line commit message.
Yes yes a Dropbox comment. But the problem here is 1 million people are doing the same thing. For this to be worth 60M seed I suspect they need to do something more than you can achieve by messing around locally."
"Claude build me a script in bash to implement a Ralph loop with a KV store tied to my git commits for agent memory."
It is not the system that is on crack ...
I think the moat depends on how long it takes for an agent to ingest the entire commit history and product documentation into context on the fly. At the rate models are improving, seeing the reasoning chain of an outdated model that led to a commit that warrants post-hoc review (likely becuase of a bug) would mainly be useful for root cause analysis more than for insight into what to do next... but chances are the newer model would have been able to infer it from local context anyway.
It also creates a challenge with respect with the embedding model chosen and how future proof it turns out to be.
I landed on a similar vision last year. The more I thought about it, the moat felt fragile. GitHub or GitLab could build the same capabilities and become a natural extension of what teams already use. That said, it addresses a real problem, and the SDLC needs to evolve.
Agents can save their reasoning into markdown files, and commit those files to Git. Are "Checkpoints" just a marketing term for that, or there's more to it?
Claude Code already does this, you can access it with /resume, /rewind and /fork. I'd imagine building a version that saves in the repo instead of in the home folder would take very minimal effort.
This is about doing it seamlessly and flawlessly then sharing it across a team.
So using something like the compound engineering plugin and committing its "brainstorms", plans, and "solutions"?
No, it's not. I can't get leadership's buyin on a Claude Plugin.
Hmm the whole point of checkpoints seems to be to reduce token waste by saving repeat thinking work. But wouldn't trying to pull N checkpoints into context of the N+1 task be MUCH more expensive? It's at odds with the current practice of clearing context regularly to save on input tokens. Even subagents (which I think are the real superpower that Claude Code has over Gemini CLI for now) by their nature get spawned with fresh near-empty context.
Token costs aside, arguably fresh context is also better at problem solving. When it was just me coding by hand, I didn't save all my intermediate thinking work anywhere: instead thinking afresh when a similar problem came up later helped in coming up with better solutions. I did occasionally save my thinking in design docs, but the equivalent to that is CLAUDE.md and similar human-reviewed markdown saved at explicit -umm- checkpoints.
60 million SEED round? This is really a thing now?
Hey, they said AI was 100x right?
Take a look at the 500M seed round next :)
Actually interesting, but how's that different from just putting your learning / decision context into the normal commit text (body) ? An LLM can search that too, and doesn't require a new cli tool.
EDIT: Or just keep a proper (technical) changelog.txt file in the repo. A lot of the "agentic/LLM engineering frameworks" boil down to best approaches and proper standards the industry should have been following decades ago.
After I have an ai dona task, I ask the next one to look at that plan and git diff and so ble check validate
I don't see the need for a full platform that is separate from where my code already lives. If I'm migrating away, it's to something like tangled, not another VC funded company
This is a good idea but I feel like you could get something similar by just adding an instruction for the agent to summarize the context for the commit into a .context/commit/<sha> file as a git hook.
Or git notes.
Commit hook > Background agent summarizes (in a data structure) the work that went into the commit > saves to a note
Built similar (with a better name) a week ago at a hackathon: https://github.com/eqtylab/y
Exactly. I don't want to wade through a whole session log just to get to reasoning, and more importantly, I don't want to taint my current agent context with a bunch of old context.
Context management is still an important human skill in working with an agent, and this makes it harder.
Feature plans generated by agents are often transient documents that fall away once the plan is executed. Ideally, that artifact would be preserved alongside the implementation.
My experience is that Cursor's reliance on VS Code's clunky panel-based UI and jack-of-all-trades harness is holding it back. Likewise, Claude Code shoe-horning a GUI into a TUI and perma-binding to a single model family is not the ideal end-state.
The VC play here? The git context CLI thing is a foundational step that lays the groundwork for a full IDE/workflow tool, I guess.
Why do you want to preserve that artifact?
If you don't have a record of questions asked/answered and rationale for decisions taken, I've noticed it's easy for subsequent feature plans to clash. Maintaining a line of consistency across each feature plan is a good thing.
>CLI to tie agent context into Git on every push.
Is this the product? I don't want to jump on the detractor wagon, but I read the post and watched the video, and all I gathered is that it dumps the context into the commit. I already do this.
> I already do this.
Hows your ability to get an enterprise to mandate their 5000 employees to use it? That's what most of these types of rounds are about.
This doesn't appear to address that concern.
I guess if I had to do it, I'd reject pushes without the requisite commit to entire/checkpoints/v1. I think their storage strategy is a bit clunky for that, but it can be done. I might look to do something more like the way jujutsu colocates its metadata. I don't think this particular implementation detail makes too much of a difference, though. I got along just fine in a regulated environment by setting a policy and enforcing it before git existed. Ideally, we'd do things for a good reason, and if you can't get along in that world, then it's probably not the right job for you. Sometimes you've got to get the change controls in order before we can mainline your contributions because we have to sustain audits. I don't think this is about forcing people to do something that they otherwise wouldn't do if you told them that it's a requirement of the job.
100%. Day one is to ship the basic capability, which many of us have already vibe-coded... Day two is all the enterprise stuff to make big companies trust AI coding more. That could unlock a lot of revenue. This isn't random at all.
Uuh easy fire them all and replace with said agents
Isnt this overloading git commits too much ? Like 50kb per commit message
Git is totally fine keeping a few extra text files. These are ephemeral anyway. The working sessions just get squashed down and eliminated by the time I've got something worth saving anyway. At that point, I might keep a overview file around describing what the change does and how it was implemented.
(I will give the agent boom a bit of credit: I write a lot more documentation now, because it's essentially instruction and initial instruction to anything else that works on it. That's a total inversion, and I think it's good.)
The bigger problem is, like others have said, there's no one true flow. I use different agents for different things. I might summarize a lot of reasoning with a cheap model to create a design document, or use a higher reasoning model to sanity check a plan, whatever. It's a lot like programming in English. I don't want my tool to be prescriptive and imposing its technical restrictions on me.
All of that aside: it's impossible that this tool raised $60 million. The problem with this post is that it's supposed to be a hype post about changing the game "entirely" but it doesn't give us a glimpse into whatever we're supposed to by hyped about.
the git commits message description never go away though, unless you're editing the git with BFG cleaner
1. Commit messages go away if you remove the commit, but
2. Don't put it in the message. Put it in files.
Same thought. If anything I'm usually trying to find ways to reduce how much context is carried over.
I have it (claude, codex) summarise what we've discussed about a design, big change, put it in an MD file and then I correct it, have it re-read it and then do the change.
Then later if it goes off piste in another session tell it to re-read the ADDs for x, y and z.
If someone could make that process less clunky, that would be great. However it's very much not just funnel every turd uttered in the prompt onto a git branch and trying a chug the lot down every session.
what about using git notes to stash the summaries? (https://git-scm.com/docs/git-notes)
Very similar for me. I have a plans folder in my root where I store the plans while they're either under improvement or under implementation. Once they're done they're moved into the plans/old folder. So far it's worked great. It's a couple of manual steps extra but very helpful record.
Pretty much the same thing. I don't find it to be a burden. Regarding the product, I'm willing to believe I just don't see big picture, but without some peek at the magic, I don't know how much easier this could really be.
Huh, the checkpoint primitive is something that I've been thinking about for a while, excited to see how it's implemented in the CLI. Git-compatible structures seem to be a pretty big pull whenever they're talking about context management.
Already at a $300 million valuation at seed. Who’s going to join this as a regular IC who gets a pittance in options? I can’t imagine the risk matches the modest potential upside.
They'll raise at double that or more before the end of the year. The dynamics of the VC market right now are staggering to watch, but the money velocity is real, and this has "ex-CEO of Github" plus "AI".
It's legit mania in VC world even as they're looking at each going "is this mania? Is it mania if you're asking if it's mania". The only rule right now is the music is playing so no-one wants to grab a chair. There's a sense this might come crashing down, but what's a player gonna do, sit on the side while paper markups are producing IRR that is practically unprecedented?
This kinda has to end badly at this point.
> this has "ex-CEO of Github" plus "AI".
If they could top it off by stealing a janitor from OpenAI or Anthropic, the VCs might wet their pants with excitement.
They'll hire one guy to vibe code like ai.com thing
I like the concept, but it seems like an obvious extension of where Anthropic is building with Tasks and Memory. What’s the moat? Cross-agent compatibility?
I would argue the largest developer tools/services acquisition in recent history was GitHub (assuming you don't count RedHat in that category). GitHub was primarily an acquisition focused on acquiring the network of developers, based on the trust and cachet GitHub had earned, with the intent to upsell (land and expand) the developers at large Enterprise companies on downstream workflows and tools.
If I were an employee looking to join Entire, or a developer evaluating the durability of Entire for my needs over the long-term, I'd ask things like —
- What's the terminal value of a DevTool in the AI era? Is it closer to a standard 10x ARR? or maybe 100x...perhaps 1000x?
- Is there value in the asset attributable to the network? If so, is it defensible? What is the likelihood protocols emerge that simply disintermediate the network moat of a AI agent memory company like Entire?
- What kind of AI data are developers willing to silo with a single vendor? What kind of controls do Enterprises expect of an AI agent memory system? Can Entire reasonably provide them to grow in the next 12-24 months?
- As a potential employee...if you join a company with a $60M seed funding and 19 employees, what is the ARR they need to achieve based on the first product in market in roughly ~12 months? $6M...$12M...20M? Is there a DevTools company that's ever done that!? What is the thesis for "this time is different"? Getting developers to pay for things is tough 'doncha know?
Only then can you ask basic technical diligence questions —
- Is Git a system that can scale to be the memory system and handle the kind of tasks that AI agents are expected handle?
- Are the infrastructure costs for these early platform decisions here reasonable over the long-term? Or do they just run out of money?
I wish them the best, and I hope employees get liquidity, and people take money off the table in a responsible way.
I don't understand how this is different from giving an agent access to github logs? The landing page is terrible at explaining what it does.I guess they are just storing context in git aswell?
So is this just a few context.md files that you tell the agent to update as you work and then push it when you are done???
I hacked together something similar to the concept they describe a few months ago (https://github.com/btucker/agentgit) and then ended up not actually finding it that useful and abandoning it.
I feel like the value would be in analyzing those rich traces with another agent to extract (failure) patterns and learnings, as part of a flywheel setup. As a human I would rarely if ever want to look at this -- I don't even have time to look at the final code itself!
> value would be in analyzing those rich traces with another agent to extract (failure) patterns and learnings
Claude Code supports hooks. This allows me to run an agent skill at the end of every agent execution to automatically determine if there were any lessons worth learning from the last session. If there were. new agent skills are automatically created or existing ones automatically updated as apporpriate.
Completely agree. But I wonder how much of that is just accomplished with well placed code comments that explain the why for future agent interactions to prevent them from misunderstanding. I have something like this in my AGENTS.md.
Try running `/insights` with Claude Code.
Checkpoints sounds like an interesting idea, and one I think we'll benefit from if they can make it useful.
I tried a similar(-ish) thing last year at https://github.com/imjasonh/cnotes (a Claude hook to write conversations to git notes) but ended up not getting much out of it. Making it integrated into the experience would have helped, I had a chrome extension to display it in the GitHub UI but even then just stopped using it eventually.
Ah you were 7mo ahead of me doing the same and also coming to a similar conclusion. The idea holds value but in practice it isnt felt.
https://github.com/eqtylab/y
What are they actually “checkpointing”? Git already “checkpoints” your code and they can’t time travel your LLM conversation.
I had a similar, admitted poorly thought out idea a few months back.
I wanted to more or less build Jira for agents and track the context there.
If I had to guess 60 million is just enough to build the POC out. I don't see how this can compete though, Open AI or Anthro could easily spin up a competitor internally.
You don't need to build anything. Just tell the agent to write tickets into .md files in a folder and move them to a closed foler as you go along. I've been using Claude Code with the Max plan nonstop essentially every day since last July and since then I've come to realize that the newer people are the more things they think they need to add to CC to get it work well.
Eventually you'll find a way that works for you that needs none of it, and if any piece of all the extras IS ever helpful, Anthropic adds it themselves within a month or two.
I'm thinking a customized LLM would write notes in its own hyper compressed language which would allow it to be much much more efficient.
For debugging you could translate it out to English, but if these agents can do stuff without humans in the loop, why do they need to take notes in English?
I can't imagine creating this without hundreds of millions if not billions. I think the future is specialized models
Isn't that basically what things like this are for, open source, free.... https://github.com/steveyegge/beads
> If I had to guess 60 million is just enough to build the POC out.
You're kidding.
In what world do you need $60M to build a POC of whatever this is supposed to be?
Shouldn't this tool be agnostic to the models? Seems like a 3rd party is the way to go.
> The game has changed. The system is cracking.
No, it hasn’t. No, it isn’t.
I miss the good old days:
https://news.ycombinator.com/item?id=338286
> Checkpoints are a new primitive that automatically captures agent context as first-class, versioned data in Git. When you commit code generated by an agent, Checkpoints capture the full session alongside the commit: the transcript, prompts, files touched, token usage, tool calls and more.
Whether or not useful for agent collaboration, the data here will be more valuable than gold for doing RL training later on.
disclosure: i run a startup that will most likely be competitive in the future.
I welcome more innovation in the code forge space but if you’re looking for an oss alternative just for tracking agent sessions with your commits you should checkout agentblame
https://github.com/mesa-dot-dev/agentblame
Entire CEO here. We are going to be building in the open and full stack open source, but great to see alternatives.
Did you have to choose an adjective to name your product. Now it’s going to be very confusing for search engines and LLms. “Tell me more about entire.” “Entire what?” “You know, that entire thing.”
Another of your competitors here. It makes me giggle that we're going after the entire developer experience while Entire is only looking at a small corner of it.
Time will tell how small that corner is. ;)
Certainly! But just to confirm, you aren't making an IDE or building a version control system to replace Git, are you? While money means you need not fear me, the scale of my vision means that I don't fear you either.
Oh I don't think I need this if all of my commits are AI!
Github but this time they're capitalizing on "agent traces". Alright. *Terrible* name btw.
I shall give the benefit of a doubt given they are "building in the open". I feel my current setup already does all this though, so I struggle to see the point
It’s funny. The whole “review intent", "learning" from past mistakes, etc, is exactly what my current set up does too. For free. Using .md files said agents generate as they go.
Soooo.. It's like Git but on the intern... Er, with agents? On agents?
My first thought that it was made for companies which tie "AI usage" to performance evaluation.
There is also Git AI: https://github.com/git-ai-project/git-ai https://usegitai.com/
love the shout but git-ai is decidedly not trying to replace the SCMs. there are teams building code review tools (commercial and internal) on top of the standard and I don't think it'll be long before GitHub, GitLab and the usual suspects start supporting it since folks the community have already been hacking it into Chrome extensions - this one got play on HN last week https://news.ycombinator.com/item?id=46871473
yep i know it's not meant to be an SCM tool but I thought it was somewhat related to what they're doing right now:
"Entire CLI hooks into your git workflow to capture AI agent sessions on every push."
Which is capturing the LLM convo along with the code (I could be wrong ofc)
This feels a bit like when some Hubbers broke off to work on PlanetScale, except without the massively successful, proven-to-be-scalable open source tool to build off (Vitess).
If you're approaching this problem-space from the ground up, there are just so many fundamental problems to solve that it seems to me that no amount of money or quality of team can increase your likelihood of arriving at enough right answers to ensure success. Pulling off something like this vision in the current red-ocean market would require dozens of brilliant ideas and hundreds of correct bets.
check out https://codecast.sh !
This sounds like Spark with extra steps. If I’m not mistaken, some version of Spark had a feature to jump on different iterations of your prompts.
I built a skill for this: https://github.com/doubleuuser/rlm-workflow
The readme is a bit more to the point.
Just have a data lake with annotated agent sessions and tool blobs (you should already be keeping this stuff for evals), then give your agent the ability to query it. No need for a special platform, or SaaS.
As for SDLC, you can do some good automations if you're very opinionated, but people have diverse tastes in the way they want to work, so it becomes a market selection thing.
This is the way
Its a shame Pierre shut down. Wish they could have made it work. Github but made by Linear would be a dream.
Pierre didn't shutdown, they said they just paused signups on the code review app to focus on the code storage service.
Productizing the building blocks of the platform seems like the smart play in today's environment honestly.
Sure but I dont want to build my own Github I just want to use a beautiful and faster alternative
We’re shooting for this with Tangled (https://tangled.org). :)
I'm interested to see if they will try to tackle the segregation of human vs AI code. The downside of agents is that they make too much changes to review, I prefer being able to track which changes I wrote or validated from the code the AI wrote.
Ironically, I was shortly contracting on PoC similar to this for ex github cofunder around this time last year.
For people trying to understand the product (so far), it seems that entire is essentially an implementation of the idea documented by http://agent-trace.dev.
Really hope that unlike GitHub it'll be open source
With a 60M seed? I doubt it.
> Cursor's Composer 2.0
There is no Composer 2.0. There is Cursor 2.0 and Composer 1.5.
Highly dubious of this.
I see zero reason for a person to care about the checkpoints.
And for agents, full sessions just needlessly fill context.
So not sure what is being solved by this.
`/init` is good enough.
Model improvements will take care of the rest.
There have been so many GitHub CEOs I was excited to find out which one.
Only four: Chris, Tom, Nat, and Thomas. Last one is me. ;)
PJ was technically CEO for awhile when they needed someone to do it
I do not want it.
What's the long-term or even short-term strategy to make money?
It's not like $60m in funding was given as charity.
General purpose agentic AI for enterprises since apparently that's the hot shit for 2026 now.
Sorry for not contributing to the discussion (as per the guidelines), but is it just me or this blog post reads a lot like LLM-filled mumble jumble? Seems like I could trim half of the words there and nothing would be lost.
I’m manually checking in Agent.md for every commit to improve the context window usage. Is that now automated?
you're doing what now?
New agent framework / platform every week now. It's crazy how fast things move...just when you get comfortable with an AI flow something new comes out...
Entire.io, the name is on point considering it asks for access to my entire GitHub account.
But seriously, $300M valuation for a CLI tool that adds some metadata to Git commits. I don't know what to say.
Shoulda launched a new chat protocol to replace discord.
Can someone please explain what is this?
I am already overloaded with information (generated by AI and humans) on my day to day job, why do I need this additional context, unless company I work for just wants to spend more money to store more slop?
How is it different than reversing it, given a PR -> generate prompt based on business context relevant to the repo or mentioned issues -> preserve it as part of PR description
I barely look at git commit history, why should I look for even higher cardinality data, in this case: WTF, are you doing, idiot, I said don't change the logic to make tests pass, I said properly write tests!
I see the vision here. I think this is extremely needed.
check out codecast.sh !
> ... to Cursor's Composer 2.0 and more, ...
I couldn't find any references of Composer 2.0 anywhere. When did that come out?
1.5 released yesterday. probably just slop
- https://cursor.com/blog/composer-1-5
Fixed. It's Cursor 2.0 and Composer 1.5, mixed that up when editing the post last night.
I bet it will down/unstable 3/4 of the month.
I love this. Prompt engineering is a real skill, and learning by example is the best way to do it. Juniors being able to see how seniors actually did something is really powerful.
I don't see how we need a brand new paradigm just because LLMs evidently suck at sharing context in their Git commits. The rules for good commits still apply in The New Age. Git is still good enough, LLMs (i.e. their developer handlers) just need to leverage it.
Personally, I don't let LLMs commit directly. I git add -p and write my own commit messages -- with additional context where required -- because at the end of the day, I'm responsible for the code. If something's unclear or lacks context, it's my fault, not the robot's.
But I would like to see a better GitHub, so maybe they will end up there.
With the how AI companies are advertising we can just tell the AI what we want and it will be done with no additional human interaction needed, why do we need a new type of development platform? We shouldn't need to collaborate at all.
https://www.youtube.com/watch?v=aJUuJtGgkQg
* This is snarky. Yes. But seriously.
Wait, since when is Dohmke out? I thought this was gonna be Nat.
I left August 11: https://github.blog/news-insights/company-news/goodbye-githu...
I’m team Geoffrey Huntley
[flagged]
"Don't be snarky."
"Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something."
https://news.ycombinator.com/newsguidelines.html
With openclaw we won't need to make event those apps.
Essentially all software is augmented with agentic development now, or if not, built with technology or on platforms that is
It's like complaining about the availability of the printing press because it proliferated tabloid production, while preferring beautifully hand-crafted tomes. It's reactively trendy to hate on it because of the vulgar production it enables and to elevate the artisanal extremes that escape its apparent influence
It's really not as integral as you make it sound. If I make one PR on a widely used open source tool with a small fix, is most software development augmented by me?
Outside of simply not being true, the sentiment of what you're saying isn't much different than:
"Essentially all software is augmented with Stack Overflow now, or if not, built with technology or on platforms that is."
Agentic development isn't a panacea nor as widespread as you claim. I'd wager that the vast majority of developers treat AI is a more specified search engine to point them in the direction they're looking for.
AI hallucination is still as massive problem. Can't tell you the number of times I've used agentic prompting with a top model that writes code for a package based on the wrong version number or flat out invents functionality that doesn't exist.
I just cannot fathom how people can say something like this today, agentic tools have now passed an inflection point. People want to point out the short comings and fully ignore that you can now make a fully functioning iPhone app in a day without knowing swift or front end development? That I can at my company do two projects simultaneously, both of them done in about 1/4 the time and one would not have even been attempted before due to the SWE headcount you would have to steal. There are countless examples I have in my own personal projects that just are such an obvious counter example to the moaning “I appreciate the craft” people or “yea this will never work because people still have to read the code” (today sure and this is now made more manageable by good quality agents, tomorrow no. No you won’t need to read code.)
I've found that the effort required to get a good outcome is roughly equal to the effort of doing it myself.
If I do it myself, I get the added bonus of actually understanding what the code is doing, which makes debugging any issues down the line way easier. It's also in generally better for teams b/c you can ask the 'owner' of a part of the codebase what their intuition is on an issue (trying to have AI fill in for this purpose has been underwhelming for me so far).
Trying to maintain a vibecoded codebase essentially involves spelunking though a non-familliar codebase every time manual action is needed to fix an issue (including reviewing/verifying the output of an AI tool's fix for the issue).
(For small/pinpointed things, it has been very good. e.g.: write a python script to comb through this CSV and print x details about it/turn this into a dashboard)
In sonnet 4 and even 4.5 I would have said you are absolutely right, and in many cases it slows you down especially when you don’t know enough to sniff trouble.
Opus 4.5 and 4.6 is where those instances have gone down, waaay down (though still true). Two personal projects I had abandoned after sonnet built a large pile of semi working cruft it couldn’t quite reason about, opus 4.6 does it in almost one shot.
You are right about learning but consider: you can educate yourself along the way — in some cases it’s no substitute for writing the code yourself, and in many cases you learn a ton more because it’s an excellent teacher and you can try out ideas to see which work best or get feedback on them. I feel I have learned a TON about the space though unlike when I code it myself I may not be extremely comfortable with the details. I would argue we are about 30% of the way to the point where it’s not even no longer relevant it’s a disservice to your company to be writing things yourself.
I didn’t say essentially all software is vibe coded. You already agree with me that it’s very good at some range of common tasks.
There are other things very good "at some range of common tasks". For example, stackoverflow snippets, libraries, bash spaghetti and even some no-code/low-code tools.
ok
What part of Voyager I and Voyager II are "augmented with agentic development?"
Surely if all software is augmented with agentic development now, our most important space probes have had their software augmented too, right?
What about my blog that I serve static pages on? What about the xray machine my dentist uses? What about the firmware in my toaster? Does the New York Stock Exchange use AI to action stock trades? What about my telescope's ACSOM driver?
You’re talking about a 1970s satellite? I guess you win the argument?
Blog: I use AI to make and blog developers are using agentic tools
X-ray machine: again a little late here, plus if you want to start dragging in places that likely have a huge amount of beaurocracy I don’t know that that’s very fair
Firmware in your toaster: cmon these are old basic things, if it’s new firmware maybe? But probably not? These are not strong examples
NYSE to action on stock trades; no they don’t use AI to action on stock trades (that would be dumb and slow and horribly inefficient and non-deterministic), but may very well now be using AI to work on the codebase that does
Let’s try to find maybe more impactful examples than small embodied components in toasters and telescopes, 1970s era telescopes that are already past our solar system.
The denial runs deep
So you admit that AI isn't in every software, and yet somehow I'm the one in denial?
Im saying you’re missing the point and the spirit of the argument. Yes, you are right, voyager doesn’t use agentic AI! I don’t even think the other examples you used are as agentic free as you think. They may or may not be! What’s the point you want to make?
Huh? Software under development obviously not software made before these tools existed
I am not willing to share my sheepish prompts with my team. Sorry!
Hah. "If it's not too much trouble, would you mind if we disable the rimraf root feature?"
Gotta bully that thing man. There's probably room in the market for a local tool that strips the superfluous niceties from instructions. Probably gonna save a material amount of tokens in aggregate.
I'm with you. I start every new prompt with: "Good morning", even at midnight. I'll be so embarrassed if that leaks.
LOL
Not sure what it is or what it does.
Uses AI to summarize coding sessions tied to commits.
Commit hook > Background agent summarizes (in a data structure) the work that went into the commit.
Built similar (with a better name) a week ago at a hackathon: https://github.com/eqtylab/y
Which only reinforces someone just lit $60M on fire. It's trivial to do this and there are so many ways people do things, having the AI build custom for you is better than paying some VC funded platform to build something for the average
Not even pocket change compared to the billions of VC money burnt every month to keep the show running.
It extracts money from investors and allocates it to founders.
Yay, MORE 'AI' agents! Hint: There are already too many Artificial Indians!
The lack of explanation of what it is and does is a tell of what gullible audience they are seeking.
Tech marketing has become a lot like dating, no technical explanation and intellectual honesty, just word words words and unreasonable expectations.
People usually cannot be honest in their romantic affairs, and here it is the same. Nobody can state: we just want to be between you and whatever you want to accomplish, rent seeking forever!
Will they ever care to elaborate HOW things works and the rationale behind stating this provides any benefit whatsoever? Perhaps this is not intended for those type of humans that care about understanding and logic?
Looking at the CLI implementation. Why not build on top of jj?
most people use git, jj has compatibility gaps
Which CEO?
$60M seed to wrap git hooks in YAML config. The AI tooling bubble is just VCs subsidizing solutions looking for problems while developers want less complexity, not more.
I don't want agent context tied to git commits. I just want infinite scroll in Claude Code and ability to search and review all my past conversations!
$300kk valuation for git commits :) the bubble will pop at some point, I don’t know when, but boy will it be spectacular.
Sounds very cringe
Not surprising for a $60M seed round
Do we have new words for smaller amounts or is this inflation at work?
I really hate this trend of naming companies using dictionary words just because they can afford to spend cash on the domain name instead of engineering. Render, fly, modal, entire and so on.
Clicks through to see what Tom or Chris started…
Oh, nevermind, it’s some MS dude.
I did test it and use it and trashed it because there is very little value, actually none for me. These problems are easily being solved in other ways whoever has any experience with these tools. Getting $60M round for this stuff is ridiculous.
so github ci/cd agents rebranded as a startup? same team different company.
Christ, a $60m seed round.
The AI fatigue is real, and the cooling-off period is going to hurt. We’re deep into concept overload now. Every week it’s another tool (don’t get me started on Gas Town) confidently claiming to solve… something. “Faster development”, apparently.
Unless you’re already ideologically committed to this space, I don’t see how the average engineer has the energy or motivation to even understand these tools, never mind meaningfully compare them. That’s before you factor in that many of them actively remove the parts of engineering people enjoy, while piling on yet another layer of abstraction, configuration, and cognitive load.
I’m so tired of being told we’re in yet another “paradigm shift”. Tools like Codex can be useful in small doses, but the moment it turns into a sprawling ecosystem of prompts, agents, workflows, and magical thinking, it stops feeling like leverage and starts feeling like self-inflicted complexity.
It's an ex-CEO of Github. He can raise $60m on any idea.
> I don’t see how the average engineer has the energy or motivation to even understand these tools, never mind meaningfully compare them
This is why I use the copilot extension in VS code. They seem to just copy whatever useful thing climbs to the surface of the AI tool slop pile. Last week I loaded up and Opus 4.6 was there ready to use. Yesterday I found it has a new Claude tool built in which I used to do some refactoring... it worked fine. It's like having an AI tool curator.
Maybe just learning 1 or 2 of such tools is enough ?
Probably, but which ones, do we get to a place where you have X years experience in Gastown development, but I only have Y years experience in Entire.
I also keep getting job applications for AI-native 'developers' whatever that means.
You will learn a lot about the underlying LLM / technology whichever tool you use though
History has shown that by delaying learning the next greatest tech, you may avoid learning it altogether.
Your point about the overwhelming proliferation of AI tools and not knowing which are worth any attention and which are trash is very true I feel that a lot today (my solution is basically to just lean into one or two and ask for recommendations on other tools with mixed success).
The “I’m so tired of being told we’re in another paradigm shift” comments are widely heard and upvoted on HN and are just so hard to comprehend today. They are not seeing the writing on the wall and following where the ball is going to be even in 6-12 months. We have scaling laws, multiple METR benchmarks, internal and external evals of a variety of flavors.
“Tools like codex can be useful in small doses” the best and most prestigious engineers I know inside and outside my company do not code virtually at all. I’m not one of them but I also do not code at all whatsoever. Agents are sufficiently powerful to justify and explain themselves and walk you through as much of the code as you want them to.
Yeah, I’m not disputing that AI-assisted engineering is a real shift. It obviously is.
My issue is that we’ve now got a million secondary “paradigm shifts” layered on top: agent frameworks, orchestration patterns, prompt DSLs, eval harnesses, routing, memory, tool calling, “autonomous” workflows… all presented like you’re behind if you’re not constantly replatforming your brain.
Even if the end-state is “engineers code less”, the near-term reality for most engineers is still: deliver software, support customers, handle incidents, and now also become competent evaluators of rapidly changing bot stacks. That cognitive tax is brutal.
So yes, follow where the ball is going. I am. I’m just not pretending the current proliferation is anything other than noisy and expensive to keep up with.
"$60M Seed round"
I guess when you are Ex-Github CEO, it is that easy raising a $60M seed. I wonder what the record for a seed round is. This is crazy.
I thought something got seriously wrong with Nat Friedman but fortunately it's another one.
Grifters to the grift god
Really struggling to figure out what this is at a glance. Buried in the text is this line which I think is the tl;dr:
"As a result, every change can now be traced back not only to a diff, but to the reasoning that produced it."
This is a good idea, but I just don't see how you build an entire platform around this. This feels like a feature that should be added to GitHub. Something to see in the existing PR workflow. Why do I want to go to a separate developer platform to look at this information?
I'm sure i'm missing something but can you not ask the llm to add the reasoning behind the commit in the comments as part of the general llm instructions?
Github sucks now, for one; people are looking for an alternative.
This is not an alternative to GitHub though. The code for this tool itself lives on GitHub!
https://github.com/entireio
Oh man I'm tired. This reminds me of the docker era. It's all moving fast. Everyone's raising money. And 24 months from now it's all consolidating. It's all a nice hype game when you raise the funding but the execution depends on people finding value in your products and tools. I would argue yes many of these things are useful but I'd also argue there's far too much overlap, too many unknowns and too many people trying to reinvent the whole process. And just like the container era I think we're going to see a real race to zero. Where most of the dev tools get open sourced and only a handful of product companies survive, if that. I want to wish everyone the best of luck because I myself have raised money and spent countless years building Dev tools. This is no easy task especially as the landscape is changing. I just think when you raise $60m and announce a cli. You're already dead, you just don't know it. I'm sorry.
Let the cambrian explosion run its course but let's hope the meteorite doesn't kill us all.
I see the value since I built a similar tool different approach. Then there's Beads, which is what inspired my project, with some tens of thousands of developers using it or more now? I'm not sure how they figure how many users they have.
In my case I don't want my tools to assume git, my tools should work whether I open SVN, TFS, Git, or a zip file. It should also sync back into my 'human' tooling, which is what I do currently. Still working on it, but its also free, just like Beads.
I wouldn't wanna be in the rat race myself, but I know people who salivate at the opportunity to create some popular dev tool to get acquired by MS, Google or Amazon or whichever of the big tech companies that decide this could work well in their cloud ecosystem.
HNites are hilarious.
On the one hand they think these things provide 1337x productivity gains, can be run autonomously, and will one day lead to "the first 1 person billion dollar company".
And in complete cognitive dissonance also somehow still have fantasies of future 'acquisition' by their oppressors.
Why acquire your trash dev tool?
They'll just have the agents copy it. Hell, you could even outright steal it, because apparently laundering any licensing issues through LLMs short circuits the brains of judges to protohuman clacking rocks together levels.
I think there are 2 parts here. That persona you’re describing (startup cofounder or engineer being paid mostly in equity) is a good subset of the people here. If I had to pull a number out of my shiny metal ass, I’d say it’s 30%. Those people both loath big tech, and dream of the day they are acquired by it. It’s not really the contradiction you think it’s. Another 45% of people here are tech-savvy Reddit refuges who say Reddit things.
As to why would those company acquire a startup instead of having an agent generate it for them. Why has big tech ever acquired tech startups when they could have always funded it in house? It’s not always a technical answer. Sometimes it’s internal Political fights, time to market, reduce competition, PR reasons or they just wanna hire the founder to lead a team for that internally and the only way he’ll agree is if there is an exit plan for his employees. I sat in “acquire or build” discussions before. The “how hard would it be to just do that?” Was just one of many inputs into the discussion. Ever wondered why big big companies acquire a smaller one, not invest in it, then shut it down few years later?
What if it's just the beginning of something bigger?
What if the earth exploded tomorrow? Who cares about what if.
With 60 million you could have waited for a bigger announcement? There's "AI fatigue" among the target market for these sorts of tools, advertising unfinished products will take its toll on you later.
List of Github CEOs:
1. Tom Preston-Werner (Co-founder). 2008 – 2014 (Out for, eh... look it up)
2. Chris Wanstrath (Co-founder). 2014 – 2018
(2018: Acquisition by Microsoft: https://news.ycombinator.com/item?id=17227286)
3. Nat Friedman (Gnome/Ximian/Microsoft). 2018 – 2021
4. Thomas Dohmke (Founder of HockeyApp, some A/B testing thing, acquired by Microsoft in 2014). 2021 - 2025
There is no Github CEO now, it's just a team/org in Microsoft. (https://mrshu.github.io/github-statuses/)
Chris was also CEO from 2008 to 2012. Tom had 2012 to 2014.
Nat's company Xamarin was acquired by Microsoft in 2016.
HockeyApp wasn't A/B testing, but a platform for iPhone, Mac, Android, and Windows Phone developers to distribute their beta version (like what TestFlight is today to the App Store), collect crash reports (like what Sentry is today), user feedback, and basic analytics for developers.
Thanks for the fact check :).
The Ximian thing I wrote from obviously faulty memory (I now wonder if it was influenced by early 2000s Miguel's bonobo obsession), the rest from various google searches. Should have gone deeper.
No, I was actually correct.
https://en.wikipedia.org/wiki/Ximian
Ximian, Inc. (previously called Helix Code and originally named International Gnome Support) was an American company that developed, sold and supported application software for Linux and Unix based on the GNOME platform. It was founded by Miguel de Icaza and Nat Friedman in 1999 and was bought by Novell in 2003
...
Novell was in turn acquired by The Attachmate Group on 27 April 2011. In May 2011 The Attachmate Group laid off all its US staff working on Mono, which included De Icaza. He and Friedman then founded Xamarin on 16 May 2011, a new company to continue the development of Mono. On 24 February 2016, Microsoft announced that they had signed an agreement to acquire Xamarin.
Didn't say you were wrong. It was just missing the Xamarin step in the sequence of companies, and arguably Xamarin was the bigger milestone than Ximian.
We went from having new JavaScript frameworks every week to having new AI frameworks every week. I'm thinking I should build a HN clone that filters out all posts about AI topics...
Looking at the most popular agent skills, heavily geared towards react and JS, I think a lot of the most breathless reports of LLM success are weighted towards the same group of fashion-dependant JavaScript developers.
The same very online group endlessly hyping messy techs and frontend JS frameworks, oblivious to the Facebook and Google sized mechanics driving said frameworks, are now 100x-ing themselves with things like “specs” and “tests” and dreaming big about type systems and compilers we’ve had for decades.
I don’t wanna say this cycle is us watching Node jockies discover systems programming in slow motion through LLMs, but it feels like that sometimes.
Ironically, using LLM’s for React is an exercise in pain, because they’re all trained on the lowest common denominator. So even Opus is constantly fighting stupid reactivity bugs.
Create extension that does that. AI can do that for you in 10 minutes
Or, you could perform a public service by creating a HN clone only for bots and try to convince the bots trolling here to go there.
That actually (sorta) already exists: https://news.ysimulator.run/news
You know the only effective way to do that, right?
Claude create a clone of Hacker News, no mistakes. Must compile.
Just give me your bank account, claude API, Mother's maiden name, your zip code, your 3 digit security code, and anything else you think I might need to live as malfist the magnificant. Can I call you that?
Yep exactly, a Perl script
I've long wished for a 'filter' feature for the hn feed -- namely the old trend of web3 slop -- but with little else than keywords to filter, it would likely be tedious and inaccurate. Ironically, I think with AI/LLMs it could be a little easier to analyze.
one technique i've found useful is i don't click on the link if i'm not interested.
it's very effective.
and there's even a "hide" link.
It’s one reason I hoped lobste.rs had taken off. All posts are tagged and you can filter out by tag.
This is how software is being written now. What you propose is like joining a forum called "Small-Scale Manufacturing News" and filtering out all 3D-printing articles.
We want to filter out the irrelevant software :)
This sounds like a company idea someone just came up with yesterday off the cuff, pitched it, and got money for because of their credentials so no one can really say no to investing in it, despite nothing new or different? What's the service or product and how is it different than every 3rd Show HN?
Either the models are good and this sort of platform gets swept away, or they aren’t, and this sort of platform gets swept away.
The most interesting thing about everyone trying to position themselves as AI experts is the futility of it: the technology explicitly promises tomorrows models will be better then todays, which means the skill investment is deflationary: the best time to learn anything is tomorrow when a better model will be better at doing the same work - because you don't need to be (conversely if you're not good at debugging and reverse engineering now...)
the best time to learn anything is tomorrow when a better model will be better at doing the same work
doesn’t that presume no value is being delivered by current models?
I can understand applying this logic to building a startup that solves today’s ai shortcomings… but value delivered today is still valuable even if it becomes more effective tomorrow.
I think it also presumes that the skills of today won't be helpful in making you better, faster, stronger at knowing what to learn tomorrow. Skateboarding ain't snowboarding but I guarantee the experience helps.
Yeah but neither makes a difference to taking a taxi.
And your skills at catching a cab don't matter for booking a self driving car online.
I'm pretty much just rawdogging Claude Code and opencode and I haven't bothered setting up skills or MCPs except for one that talks to Jira and Confluence. I just don't see the point when I'm perfectly capable of writing a detailed prompt with all my expectations.
The problem is that so many of these things are AI instructing AI and my trust rating for vibe coded tools is zero. It's become a point of pride for the human to be taken out of the loop, and the one thing that isn't recorded is the transcript that produced the slop.
I mean, you have the creator of openclaw saying he doesn't read code at all, he just generates it. That is not software engineering or development, it's brogrammer trash.
I think the rationale is that with the right tools you can move much faster, and not burn everything to the ground, than just rawdogging Claude. If you haven't bothered setting up extra tools you may still be faster / better than old you, but not better than the you that could be. I'm not preaching, that's just the idea.
> That is not software engineering or development, it's brogrammer trash.
Yes, but it's working. I'm still reading the code and calling out specific issues to Claude, but it's less and less.
That’s true for “tips and tricks” knowledge like “which model is best today” or “tell the model you’ll get fired if the answer is wrong to increase accuracy” that pops up on Twitter/X. It’s fleeting, makes people feel like “experts”, and doesn’t age well.
On the other hand, deeply understanding how models work and where they fall short, how to set up, organize, and maintain context, and which tools and workflows support that tends to last much longer. When something like the “Ralph loop” blows up on social media (and dies just as fast), the interesting question is: what problem was it trying to solve, and how did it do it differently from alternatives? Thinking through those problems is like training a muscle, and that muscle stays useful even as the underlying technology evolves.
It does seem like things are moving very quickly even deeper than what you are saying. Less than a year ago langchain, model fine tuning and RAG were the cutting edge and the “thing to do”.
Now because of models improving, context sizes getting bigger, and commercial offerings improving I hardly hear about them.
> what problem was it trying to solve, and how did it do it differently from alternatives?
Sounds to me like accidental complexity. The essential problem is to write good code for the computer to do it's task?
There's an issue if you're (general you) more focused on fixing the tool than on the primary problem, especially when you don't know if the tool is even suitable,
You nailed it. Thats exactly how I feel. Wake me up when the dust settles, and i'll deep dive and learn all the ins and outs. The churn is just too exhausting.
You might wake up in a whole different biome, Rip Van Winkle.
I don't get the pressure. I don't know about you, but my job for a long time has been continually learning new systems. I don't get how so many of my peers fall into this head trip where they think they are gonna get left behind by what amounts to anticipated new features from some SaaS one day.
How do you both hold that the technology is so revolutionary because of its productive gains, but at the same time so esoteric that you better be ontop of everything all the time?
This stuff is all like a weird toy compared to other things I have taken the time to learn in my career, the sense of expertise people claim at all comes off to me like a guy who knows the Taco Bell secret menu, or the best set of coupons to use at Target. Its the opposite of intimidating!
I'm not scared that my skills will be obsolete, I'm scared employers will think they are. The labor market was already irrational enough as it was.
I may just be a "doomer", but my current take is we have maybe 3-5 years of decent compensation left to "extract" from our profession. Being an AI "expert" will likely extend that range slightly, but at the cost of being one of the "traitors" that helps build your own replacement (but it will happen with or without you).
> the technology explicitly promises tomorrows models will be better then todays, which means the skill investment is deflationary
This is just wrong. A) It doesn’t promise improvement B) Even if it does improve, that doesn’t say anything about skill investment. Maybe its improvements amplify human skill just as they have so far.
I have a reading list of a bunch of papers i didn't get through over the past 2 years. it is crazy how many papers on this list are completely not talked about anymore.
I kinda regret going through the SeLU paper lol back in the late 2010s.
Either the business makes a profit before it gets swept away, or it doesn't. This should be your goal: make money before your business dies. If you do that, you succeeded. Businesses are always ephemeral.
This is very well put. I think this platform can be useful but I doubt it can be something as big as the think it will be. At the end of the day it’s just storing some info with your data. I guess they are trying to be the next GitHub (and clearly have the experience :)). I doubt that success can be replicated today with this idea, even with $60 mil to burn
But think of all the investor dollars between now and then!
They know hence: forget what it does, it was created by the ex CEO of another commonly used thingy!
This guy was the ex-ceo of GitHub and can't bother to communicate his product in a single announcement post?
I am here. What did I not bother with? I wrote the blog post and it has all the details.
You’ve described a technology, not a solution to a clearly articulated problem that customers actually have. The problem that you have described is vague, and it’s unclear that it’s actually a problem at all. Finally, you don’t provide a persuasive and concrete argument about how your eventual solution—whatever that may be—will solve it.
I don’t mean to be so presumptuous as to teach Grampa how to suck eggs, but I think Amazon’s working backwards process is instructive.
Hey, is JJ compatibility in the cards? Considering the blog article hints at a goal of a developerless agent-to-agent automation platform I'm guessing developer conveniences are a side quest rn?
Yes, definitely something we are thinking of.
I am struggling to see what the details are other than high-level concepts. Perhaps a demo would be useful!
Wow, account from 2011 and just two comments, both on this article. Welcome, lurker, and good luck :)
Thanks. New startup, new approach.
[flagged]
Please don't cross into personal attack in HN comments.
You may not owe ex-Github-CEOs better, but you owe this community better if you're participating in it.
https://news.ycombinator.com/newsguidelines.html
The founder has only forked repositories on GitHub that are sort of light web development related.
His use of bombastic language in this announcement suggests that he has never personally worked on serious software. The deterioration of GitHub under his tenure is not confidence inspiring either, but that of course may have been dictated by Nadella.
If you are very generous, this is just another GitHub competitor dressed up in AI B.S. in order to get funding.
Personal attacks aren't allowed here, so I've banned the account. If you want to use HN as intended, you'd be welcome to review https://news.ycombinator.com/newsguidelines.html and then email hn@ycombinator.com with a reason to believe that you'll follow the rules in the future.
Founder here. I built commercial insurance software for Windows 95 in the 1990s, driver assistant systems at Mercedes and at Bosch in the early 2000s, dozens of iPhone apps as contractor, a startup called HockeyApp (acquired by Microsoft), and various smaller projects, mostly in Ruby on Rails. And of course, when I left Microsoft & GitHub, 10 years of green boxes were removed from my GitHub profile.
Glad to see you commenting here despite the, er, unsubstantive disposition of some of the commenters! (I'm a mod here btw.) We really don't like it when the audience responds to new things with snark and dismissiveness, and there are rules in https://news.ycombinator.com/newsguidelines.html to make that clear.
In this case I think the root problem is that the OP (https://entire.io/blog/hello-entire-world/) is the wrong genre for HN. It's a fine fundraising announcement, but that sort of enthusiastic general announcement rubs the HN audience the wrong way because what they really want is technical details. Spectacular non technical details like high valuations, etc., tend to accentuate this gap.
I mention this because if you or someone on your team wants to write a technical post about what you're building, with satisfying details and all that, then we could do a take 2 (whenever would be a good time for this).
Makes sense, and we definitely will have more technical posts coming soon and on a regular basis. And I am German, I am used to snark. :P Joking aside, thanks for the response and appreciate your work as a mod!
But ... what have you done lately?
(I jest ... kudos).
This comment is both wrong and mean-spirited.
I’ve worked with ashtom for over a decade. He’s a coding machine - easily one of the most technical executives (who ships real production code.)