// agentic security research · managed AI operations
Autonomous systems that hold up in production.
We research how AI agents fail — in the open — and we run production agents for small organizations that can't safely run them alone.
01 / RESEARCH
Agent failure modes, studied in the open
Security, reliability, and evals for agentic systems — published as lab notes and open tooling. Everything generalizable is public. Everything client-specific stays yours.
02 / MANAGED SERVICE
We run it. You run your business.
The whole stack, operated for you: domain, infrastructure, the agent itself, monitoring, spend caps, and a pager we carry. You get outcomes, not a codebase to babysit.
03 / EDUCATION
The graduation path
When you're ready to bring it in-house, we train your people on this exact stack — then hand over the keys. Leaving is a feature, not a breach.
$ ./subscribe --announcements
Occasional announcements — new projects and lab notes.
Email only. We send a confirmation email first — you're not subscribed until you click it.
✓ confirmation email sent
Check your inbox to confirm your subscription.
$ ./contact --new
Tell us what you're building_
We read every message. Expect a reply within two business days.
✓ message transmitted
Thanks — we'll be in touch shortly.