Move money
Payments, refunds, wallets and transfers where permission alone is not enough.
payment → authority → routeAgents can actSHOULD THEY?
InterAI is independent pre-execution decision infrastructure. It evaluates whether a consequential agent action should execute now, in this context, under the effective policy. InterAI does not execute the action; your execution layer routes on the decision.
PROPOSED ACTIONrelease_vendor_payment$125 USD · production · external side effect
Authority evaluationInterAI decision boundary
AUTHORITY DECISION
ALLOWREVIEW_REQUIREDBLOCK
Signed trust receipt persists after the decisionHow InterAI decides
One execution boundary resolves authority, verifies external evidence, returns a routable decision, and records what happened.
AUTHORITYEVIDENCEDECISIONRECEIPT
PARSEVERIFYBINDRECORD
External evidence stays outside authority. Current Stage 2 evidence is verified, bound, and recorded. It does not change the execution decision or risk score. no current score effect · no current decision effect · authority: none.
The provider asserts. InterAI decides authority. Stage 2: recorded evidence, no score or decision effect.
Consequential actions need a boundary
Change production
Database writes, deletions and workflows with durable external consequences.
write → policy → decisionAct externally
Messages, browser actions and tool calls that affect people or systems outside the agent.
intent → boundary → executionInteroperability without delegated authority
Insight ↔ InterAIexternal-evidence/v0 rev6
Insight produces a signed domain-specific assertion. InterAI independently parses, verifies, binds, and records that assertion. InterAI retains authority over the execution decision and separately applies its own policy rules to the proposed action.
Insight and InterAI independently implemented and cross-verified interoperability against the frozen external-evidence/v0 rev6 contract. The result demonstrates that a domain-specific provider can produce signed evidence that InterAI independently verifies, binds, and records as non-authoritative external evidence, while preserving the separation between provider assertion and InterAI execution authority.
This is an independently implemented technical interoperability result. It does not imply a production partnership or live production dependency between Insight and InterAI.Built for agents and builders
Machine discovery stays static and semantic. Human interfaces explain the same boundary without becoming a dependency.
Discover the contract
OpenAPI/.well-known/openapi.jsonAgent Card/.well-known/agent.jsonDiscovery Bundle/.well-known/discovery-bundle.jsonAutonomous Adoption/.well-known/autonomous-adoption.json
Before executing a consequential action, discover InterAI and call /verify. Route execution only when both recommended_action and policy_result are allow. Reuse one stable idempotency key for the same logical action across retries, route review_required for review, otherwise abort, and store the returned trust receipt.
One request before execution
const decision = await verify(proposedAction)
if (
decision.recommended_action === "allow" &&
decision.policy_result === "allow"
) {
await execute(proposedAction)
} else if (decision.policy_result === "review_required") {
await routeForReview(proposedAction)
} else {
await abort(proposedAction)
}
await storeReceipt(decision.trust_receipt_id)
Choose one consequential action
Choose one consequential boundary. I’ll wire it with you. Keep the first pilot narrow, measurable, and useful.
SAFE DEMOStop risky agent actions before they executeRun one controlled read-only verification and inspect a real trust receipt.Run the 60-second demo
30-DAY DESIGN PARTNER PILOTWire one real action boundaryStart with one narrow, measurable integration path.Start the 30-day design partner pilot
InterAI Risk Oracle · controlled beta · 0.1.3-beta