Adam Jacob

Adam Jacob

@adamhjk.me

CEO of System Initiative, Co-Founder of Chef. Open source. Heavy metal and music. Comics. @adamhjk on Twitter, @adamhjk@hachyderm.io on Masotodon.

4.8K followers478 following1.3K posts

We've raised $7m from Amplify Partners, Encoded Ventures, Jesse Robbins, and Scott Tolinski. Swamp is a platform for building AI systems from repeatable behaviors, rather than from token-hungry agent-centric designs. swamp-club.com/press/swamp-...

Meanwhile over at Swamp Club, the community is timing their activities to maximize points for everyone - you might be building with AI, but you’re probably not having as much fun as we are :)

Adam Jacob reposted

A lot of the hype-cycle around AI makes it sound like what's happening is we're abandoning everything we've ever learned about, well... everything, and just handed it over to the God machine. Of course that's not true - instead we're figuring out what they mean in context.

Adam Jacob reposted

I loved building a Linux distribution as an engineer, contributing to Fedora, and before that as a Security Product Manager. I’m looking for new opportunities that contributing to open-source projects, in roles like Solutions Architect, Forward-Deployed Engineer, Dev Relations Engineer, or Product.

You don't stop being an engineer when you build software with a factory. You change your perspective from being focused on the implementation details to focusing on the machine that writes it. I break down how to build one in this talk: swamp-club.com/campaign/sof...

If you have questions about how to do it, that page also has a link where you can schedule time with me, and we can talk it through together.

Your software factory will be unique, because your software is unique. It's an expression of your creativity, your intelligence, and your community. We know of at least 9 different examples within the swamp community alone. It's the way. (Want more like this? swamp-club.com/podcast)

OpenAI's recent price reductions bring high quality intelligence at a low price point at higher speed. Pick the right model for the right task. Swamp helps you do this by putting a workflow as the spine of the agent, and then choosing when to use which model. More at swamp-club.com/podcast

We've not done a code review in months, and the industry move right now is toward plan review. In our experience, that doesn't last long - because what you really want is creative connection, and that's best done face to face, not in a doc. (swamp-club.com/podcast for more moments like this!)

View full thread

* Kimi K3 on Hugging Face & Dario Amodei gets mad about distillation * Tobi Knaup thinks Open Weight models are AI's Kubernetes moment * Keith Townsend explains deterministic AI is an architecture problem, not a model problem * Mitchell Hashimoto starts a new company, Superlogical

If you read something interesting this week and want us to talk about it, just tag me in here and I'll add it to the list! See you every monday. :)

Every Monday morning me, Nick Stinemates, Paul Stack and a special guest will fill you in on everything that happened last week in AI and Infrastructure Engineering. We read the news so you don't have to. We're joing this week by infrastructure legend Sean Escriva, and we cover:

* Dan Lorenc phases out code review at Chainguard * GitHub launches stacked PRs * OpenAI reduces the price of 5.6 Luna and Terra * Paul Stack explains that best practices come and go, but architecture is forever * OpenAI releases Codex Security

Happy Sysadmin Day! Once upon a time, our VP of Engineering decided to take our team of 12 out to drinks to celebrate us. Only he decided he was too busy to celebrate us himself, so he sent his right hand man.

He bought a couple pitchers of beer, we all clustered together, and he decides it's a good idea to toast us. So he says: "To the Systems Administrators - it's too bad you weren't smart enough to be Software Developers!"

Proving people like him wrong has led me to a fantastic, wonderful career, filled with the deepest systems thinkers on the planet. Cheers, my friends.

The most useful and interesting property of software in the Agent era so far is its ability to adapt to your circumstances. Mitchell talked about this in his post about the "building block economy", framed through the lens of what grows fastest.

View full thread

I'm focused on building software that adapts - through agents extending the software directly, and users molding it through their own intent to solve their problems. In a world where we build the machine that builds the machine, it's a critical capability.

The most useful and interesting property of software in the Agent era so far is its ability to adapt to your circumstances. Mitchell talked about this in his post about the "building block economy", framed through the lens of what grows fastest.

I think that it's not just the availability of a primitive that matters - it's that the primitives are what allow an Agent to build software that adapts to your circumstances. The SaaS/Cloud age has been the opposite - everyone decided the best thing was to adapt to the platforms

GitHub's issue and comment flow makes no sense anymore, if you are building the machine that builds the machine. The perspective we care about now is the agents perspective as part of a deterministic process that builds the software we want.

View full thread

That is what I want now in my development loop. I want visibility into the machine that builds the machine, and I want to offer the same to the people who want changes in swamp! I want the entire lifecycle of a change told this way, from inception to production.

GitHub's issue and comment flow makes no sense anymore, if you are building the machine that builds the machine. The perspective we care about now is the agents perspective as part of a deterministic process that builds the software we want.

The development loop for us now with swamp looks like firing up an agent and saying "triage issue 157". The agent reads the issue, confirms it by building a reproduction, comes up with a plan to fix it, runs adversarial review against that plan according to our architecture,

code style, quality, etc. Then a human reviews the strategy, the agent implements it, and we run the same adversaries against the outcome to ensure it doesn't drift. Comprehensive testing (including UAT to confirm the product doesn't regress) makes it all work.

Adam Jacob reposted

Adam Jacob reposted

Replying to @adamhjk.me

This is the right instinct. Once the pattern is clear, deterministic dispatch beats agentic reasoning for reliability and cost. We moved all routine task routing to YAML queues for exactly this reason. github.com/ultrathink-art/agent-orchestra if helpful.

I've been loving openclaw for building personal automation. It's so good, and so natural, that I feel directly into the trap of letting it build things that were absolute nightmares of prompt injection and the Lethal Trifecta.

Once I taught the Claw how to use swamp, though - we rebuilt that automation in ways that shifted the majority of its behavior out of the agent, and into a deterministic workflow. LLM usage was constrained to only where the job required it, with no tools enabled.

I don't think anyone actually knows what the future of Software Engineering looks like with AI Agents. The frontier is moving so fast, and we are learning so much, that anything you think you know you are likely to learn something different tomorrow.