Careers
Help build a chief of staff you can trust.
We have no open roles listed. We do want to meet people who care about this problem.
What we are building
An AI that does real work inside real companies, and can be trusted to stop.
That means an agent runtime, a policy layer, memory, integrations and approvals, all designed so that safety is a property of the code and not a hope about the model.
How we work
The rules we work by.
One release at a time
We audit, build, test, document and stop. We do not reach ahead.
Tests that bite
A test must fail if the bug returns. A skipped suite counts as a failure.
Nothing untested is called working
Our docs say what is done and how it behaves.
Policy is code
The model never decides authorization. The software does, every time.
Plain language
Short sentences, concrete claims, no hype. We write for the person who has to rely on it.
Economical by habit
Deterministic code first, small models for small jobs, budgets checked before the call.
Who we would like to hear from
Engineers and operators.
If one of these sounds like you, introduce yourself. There is no job posting to apply to, but we read what we receive.
- Engineers who care about security enforced in code
- People who have built agent runtimes, tool use and orchestration
- People who write evals and care about measuring what works
- Operators who have run a messy multi-entity business and know the pain
Introduce yourself
Tell us what you have built or run, and what you would want to work on.