About ActPass

AI agents are starting to take real actions — issuing refunds, deploying code, sending email, moving money. ActPass exists for one question that gets harder as agents get more capable: should this specific action be allowed to execute, right now?

We are building the runtime authorization and evidence layer that sits between an agent and its tools. Every sensitive action gets a deterministic allow / deny / needs-approval decision, enforced by signed, short-lived Action Passports, and sealed into a tamper-evident evidence chain that an auditor can verify. The decision is deterministic by design — the same input always produces the same outcome — and it fails closed when anything is uncertain.

We are an early-stage company. Rather than publish a wall of logos we haven’t earned, we work hand-in-hand with a small number of design partners to validate the runtime authorization and evidence wedge on real agent workloads. If that sounds like a problem you have, we would like to talk.