Enterprise control plane that registers, enforces policy on, and audits every AI agent action

Cinchy was founded in Toronto in 2017 and built its reputation on the Data Collaboration Platform, a dataware approach used by banks, insurers, and government agencies to share data across systems without copying it into yet another integration layer. The company has raised roughly $25M. Its governance work now extends the same idea to AI: rather than trusting that an agent will behave, PeriMind treats the connection between an agent and an enterprise system as the thing to be registered, controlled, and recorded. That framing suits organizations that already have to prove to a regulator who touched which record and under what authority.

PeriMind works in three stages. A registry discovers AI tool endpoints and agent connections across the estate, including ones nobody declared. Runtime enforcement evaluates each action against policy before it executes, allowing, blocking, or escalating it. The audit stage keeps an immutable record of what was requested, what was decided, and what data moved, which is the evidence a compliance team needs after the fact. Because the layer sits between agents and the systems they touch rather than inside any particular model or framework, it applies uniformly across LLM providers, agent toolkits, and internal applications, and gives security, risk, and platform teams one place to see AI activity and cost.