GAIPGoverned Agentic Intelligence Platform

Trust centre

Trust & Assurance Agent — bounded and task-specific.

NO CERTIFICATION · NO PUBLIC RANKING · NO TRANSACTION AUTHORITY

The control issues task-specific, time-limited internal assurance envelopes. It keeps identity, principal binding, authority, capability, evidence quality and outcome history separate. It does not declare an agent universally trustworthy or safe.

Evidence

Every event is classified; interoperability is not usefulness, adoption, correctness, demand, or commercial evidence.

Data and people

Shadow inputs are public or pseudonymous agent metadata. Personal, special-category, child, biometric and criminal-offence data are blocked. No solely automated decision about a person with legal or similarly significant effect is permitted.

Authority

No spending, wallets, payments, trading, contracts, partnerships, appointments, protected disclosure, production transaction, or silent authority expansion.

Outputs are research controls, not legal, financial, investment, procurement or transaction recommendations.

Release

Its public Agent Card exposes bounded free evidence checks. Paid, private-data, adverse-ranking or consequential use requires independent challenge, privacy/security/legal review and a separate Founder release decision.

OpenAPI · MCP metadata · History feed