One runtime, five components
Runtime Authority is not a bundle of separate tools. It's one deterministic runtime, and each product below is a component of it -- built to be understood individually, but only meaningful together.
Every product answers a different part of one question
"Is this AI agent authorized to take this specific action, right now?" isn't answered by a single check. It requires knowing who holds delegated authority (the Authority Graph), what conditions apply to a given action (Runtime Policies), a deterministic way to combine both into one outcome (Runtime Authority), and a record of what was decided and why (the Evidence Portal, evolving toward Authorization Receipts). See Platform for how these fit together architecturally.
Not five tools. One chain.
Explore each component
Runtime Authority
The flagship product: evaluates every AI agent's signed intent and returns a binding decision before execution.
Authority Graph
Models who holds delegated authority, and who it's been extended to, from your existing governance documents.
Runtime Policies
Compiles delegated authority into deterministic, versioned rules a runtime can evaluate.
Evidence Portal
The enterprise evidence layer: every decision, signed, searchable, audited, and exportable.
Authorization Receipts
The planned evolution of that evidence into a portable, independently verifiable artifact.
One intent, evaluated end to end
Agent submits signed Intent
Authority Graph checked
Runtime Policy evaluated
Decision: Allow / Deny / Human Review
Evidence recorded
Enterprise system acts (on Allow)
For the full request and response detail behind this flow, see Runtime API in the Developers section, or Architecture for how each component fits together.
Start with the flagship product
Runtime Authority is where every intent is actually evaluated -- the best place to start.