Quorum: what happens when you actually deploy the AI-safety tooling you built
Written for the All Things Agentic Hackathon, Fortified Enterprise Fleet track.
We already build AI-safety tooling. Six separate, independently-tested deterministic tools: a prompt-injection scanner, a claim/provenance integrity engine, a structured human-deliberation protocol, a cross-session drift detector, and an audit logger. Each one worked, on its own, against the thing it was built to test. None of them had ever been asked to do the one job they were actually designed for: watch an autonomous agent, not just review a single chat message.
Quorum is what happens when you point them at that job instead.
The premise
