Author here. Yes, CC is the maintainer. When I stumble on a decent idea, I would just feed it to CC to create a pattern out of it. This was my quick and dirty approach to a public learning log with an idea that I would get back to it at some point and clean it up. Which I did on a few occasions.
See my comment above. The repository is from May when I was intensely exploring everything agentic. I used it as a public bookmarking tool and also in the hope of receiving contributions. Thanks to this HN share, I received four PRs.
Hi, author here. Honestly, I just used this as a bookmarking place for myself. Which you could infer if you go through some patterns. I’ve created a flow with CC where I would just dump a new source like a podcast, post, or whatever to have it for reference.
Treat it like a wind-tunnel test, not a wisdom oracle: the value isn’t profundity, it’s seeing what kinds of coordination and failure modes emerge when the only continuity is the record and the collaboration is the environment.
Beyond graduating students, I see model labs as “accelerators/incubators” bundling, launching, and productizing observed ideas that gain traction. The sheer strength of their platforms, the number of eyes watching them, near-zero marginal costs, and seemingly unlimited budgets mean that only slow decision-making can prevent them from becoming the next Amazons of everything.
I tend to take an open-ended sabbatical every few years. I took one in 2017 and now again in 2025. It’s not a full year off; for me, it usually lasts around three to six months. That feels like the right rhythm, given that the traditional idea of a sabbatical is taking one year off every seven years. What made the biggest difference, though, was the open-ended nature of it, not having a strict return date. A fixed six-week break wouldn’t really feel like a sabbatical to me.
The company is being rebranded, not the product. Makes total sense, considering the brand equity, and also them going in the direction of productivity suite. Could be interesting.
1. The option to leave an email for those who are interested is a huge plus, obviously without forcing it in any way. Maybe I should word it differently.
2. I’m not suggesting just slapping logos on the site, which has become the norm these days even if someone only tried the product once. But seeing an endorsement from someone like Mitchell Hashimoto on ampcode.com is a plus for me.
3. Yes, “visible founders” can absolutely backfire if it becomes performative or ego-driven. But handled with humility and substance, founder presence can actually accelerate trust-building.
This came from watching a lot of promising devtools get early traction, then stall when trust signals weren’t consistent. Would love to hear how others think about earning developer trust without over-marketing it.
Exactly this. I’ve spent some time last week at a 50 something people web agency helping them setup QA process where agents explore the paths and based on those passes write automated scripts that humans verify and put into testing flow.
For whatever you think, this is well written and equipped piece of writing. But on the topic at hand, one could also think this whole drama is orchestrated by marketing. :)
This whole article reads like pre AI thinking. Mentinoning briefly AI in intro is not enough. IMO the really interesting issue, considering our AI future, won’t be discovering your values through introspection, it's figuring out how to translate human values into mathematical constraints that don't immediately get gamed by optimizers. On the other hand, core validator questions are actually pretty solid, especially the "does the inverse strike a nerve" test.
Nice to see this out there! Been playing with the token efficiency, looks like around 10% savings vs regular JSON/YAML. Not massive but it adds up. The embedded code thing is cool. Tired of configs that need weird templating or scripts all over the place just to do basic logic. Will check this out on some projects.
The process was confusing. It took for sure at least 100 sessions, haven’t checked the logs, was running it in a VM. And every time it tries to build something while forgetting all of it and then reconstructing bits and pieces just to continue. The article was written at least over 3 context lengths. But it does reconstructs well the iteration on the side of an AI agent.
No reason, all fine. Honestly, it is very hard to find time to write anything down and imagine 15k words deep analysis of the process like this. And LLMs are ideal log keepers. I also did bunch of similar experiments like doing a research paper from data to code and writing. We can only expect things to get better from here.
Sorry, will move it up. For me it was, hey let’s loop it once more over the repo and let it write about it, more like an archeological dig to unearth the process, as I wasn’t involved in it, especially not algo decisions and later optimizations.
That should be moved toward the top, will do it manually. This was 98% loop, albeit very messy one. More an exploration of the process itself. The biggest value is the meta learning. Ideally we should save traces of prompts and process itself as a verifiable or observable artifact, instead of code itself. At the end of the day, outcome over code.
Just do the stuff in the same space, either by aligning your current work or shipping hobby projects, writings, opinions on social media. You will quickly “map” the people in the area you are interested in and your only task is getting on their radar. Then you can expect them to reach out. This solves all the tediousness of cold outreach or interviews, etc. Worked well for me for the last two decades.
Atuin has been a part of my dotfiles install.sh for the last two years. It became such an integral part of my flow that it became invisible. Prob not a good thing for them, but also a testament to how naturally it fits in.