ThirtySignals is an AI engineering company.
AI is creating a new computing paradigm. That paradigm requires a new engineering discipline. We exist to help define and build it, through production systems, research, open source, and products.
Why we exist
The capability arrived faster than the discipline.
Teams are shipping systems that reason, act, and remember, using engineering practices built for systems that only executed instructions. Some of it works. A lot of it is held together by hope, duct tape and a demo that went well once.
We think this next generation of software needs a different standard. The same kind of attention that infrastructure gets: less focus on what a system can do in a demo, more on whether it can be trusted with real work, run by real people, with real consequences when it's wrong.
What we do
How we put it into practice.
- 01
We build production AI systems for enterprises.
Systems that create measurable business value and run every day — with the data pipelines, orchestration, evaluation, access control, and failure handling backed by engineering discipline that makes them trustworthy.
- 02
We build our own products and open source.
We believe the best AI engineering companies are also product companies. Building our own platforms, open-source projects, and AI-native products forces us to solve the same reliability, scalability, security, and operational challenges our clients face.
- 03
We publish what we learn.
Our research, tooling, and engineering practices go out in public — including the parts that didn't work. Closed systems ask for trust. Open ones earn it.
Products and open source
What we build.
The Iris Stack is our infrastructure for trustworthy AI agents: Iris Core runs durable work across models and tools; RelayKit puts human approval and an audit trail into the flow; Pupil continuously evaluates the system and catches regressions. It exists because of the things that actually break AI systems in production — agents dying between sessions, no audit trail for agent actions, model lock-in — not because of what looks good in a demo.
It's open source under Apache 2.0, and it's the system our own engineering practices get tested against first.
Explore the Iris Stackiris-core on GitHub(opens in a new tab)
Charter
What we believe.
The Charter is where we write down our principles about building AI systems, and version it so you can see how our thinking evolves. Three of the thirty principals:
Writing
Recent notes.
12 Aug 2026
From Chatbot to Coworker: What It Took to Put an AI Agent Into Production(opens in a new tab)
Lessons from moving Zara, a travel-curation agent built on iris-core, from a Slack prototype to a production workflow that people can trust with real clients.
ThirtySignals on Substack
3 Aug 2026
Hardening LLM Agents with Shell Access(opens in a new tab)
A production security architecture for shell-capable agents: observability, secrets management, and credential injection each solve a distinct problem.
ThirtySignals on Substack
17 Jul 2026
Vercel Shipped Eve.dev - We Think It Matters(opens in a new tab)
Why durable agent harnesses need more than prompt and tool calling: sessions, channels, skills, sandboxing, background work, and clear failure handling.
ThirtySignals on Substack
If you're building this, we'd like to hear about it.
We're most useful to teams already running AI systems in production and trying to figure out what "trustworthy" means for the ones they haven't shipped yet.