An autonomous software organization for any product

Connect a website, a mobile app or a package, and thirteen agents run its lifecycle — development, testing, monitoring, improvement, support — 24/7, stopping only at the five gates a human owns.

Create an account Sign in

It starts in paper mode: every integration is mocked and nothing reaches a real system.

The organization

Thirteen roles, in the order work moves through them. Each has its own model tier, its own daily spending cap, and only the capabilities its role file grants.

Product
Research
Architect
Coding
Code review
Test
Security
Release
SRE
Analytics
Support
Growth
Finance

Where it stops

Five decisions belong to a person, and agents work up to one and wait. Rollback, halting a rollout and turning a flag off never wait — undoing damage should never need a signature.

⚑ Merge
a change reaching trunk
⚑ Deploy
a release reaching production
⚑ Spend
a role or product that has hit its cap
⚑ Customer-facing
anything an agent wrote being sent or published
⚑ Legal
a change to terms, privacy, pricing or licensing

Any stage of life

Running in production, half-built or maintenance-only. Missing tests lower how deep agents may go and are shown as the reason, rather than blocking anything.

Any language, at one of three depths

A pack claims what it recognises; everything else gets the generic pack, capped at the shallowest level and never deploying. What differs is depth, not acceptance.

Commands are proved, not guessed

Discovery runs a component's build and test commands in a sandbox before writing them down. A missing toolchain is reported as missing rather than as a failure.