How we work
Forward-deployed, from first agent to production.
We don't hand you a dashboard and wish you luck. compileHQ embeds with your team, gets one agent governed in your infrastructure, and proves the control plane, then expands from there.
The engagement
One agent, governed end to end.
- 01
Pick where to start
One use case: an agent you already run, or one you've held back because you couldn't govern it. One thing governed end to end beats a broad rollout that enforces nothing.
- 02
Author, then compile
Your team authors the playbook, the agent's intent and policy. compileHQ compiles it into a least-privilege skill: the exact tools and segment it's allowed to touch, reviewed and signed off by security.
- 03
Connect the gateway
compileHQ goes in front of your MCP servers, embedded or as a front-proxy, inside your infrastructure. It reads the tool surface and applies your skills.
- 04
Enforce at every call
Agents authorize just-in-time and run scoped to their segment. Allows and denials are sealed into a hash-chained audit trace your auditors can export.
- 05
Expand the control plane
With one agent under control, the rest follow: more skills, more segments, more agents, all governed by the same enforcement layer.
Pricing
Priced to the work, not to a seat count.
compileHQ is a forward-deployed engagement, so pricing is scoped to it: the agents we govern, the MCP servers we put behind it, and the infrastructure we deploy into. There is no per-seat list and no tier to size yourself into.
We'll talk through where agents fit, govern the first one, and put a number against that work. Your skills stay customer-owned.
Talk to us
Run agents you can actually trust.
Grab a time that works, or leave your email and we'll reach out. Either way it's a working session, not a sales call.