A team of builders.

Conduct AI is built by engineers who ran cloud infrastructure and observability teams for the better part of a decade. Before this we built and ran Xervmon, a cloud management platform, the kind of product that taught us exactly where governance tools break under pressure, where audit trails go quiet at the moment they're needed, and where the gap between a written policy and an enforced policy can swallow a company whole.

Xervmon didn't survive. The market wasn't there yet. What did survive was a very specific understanding of how engineering teams actually make decisions when things break, and what kind of safety net is, and isn't, there when they do.

When AI arrived as a real shift in what a small team could build, we started shipping with it. Claude Code, MCP servers, agentic playbooks. The pattern was familiar: powerful tools, invisible risks, no clear owner for the layer underneath. Credentials drifting into prompts, policies that existed only on paper, no answer to the compliance question the auditor would ask in six months.

That's why Conduct AI exists.

We build the governance layer at the same layer as the tools — not above them in a policy document, not below them in infrastructure no one configures. Guard policies enforced in every AI session. Security checks in every PR. Workflows that inherit your rules instead of routing around them. An audit trail that answers the compliance question in one click, not one quarter.

Along the way we built infrastructure to make AI work better at the engineering layer. These tools are free and open source — they fund and dogfood the platform.

Source on GitHub, free for individuals and small teams. No demo gates before you can see what you're buying.

If you're building in this space and want to compare notes, we're reachable below.