Video / Evidence / PactVerity technical series

The AI Agent Trust Stack: Identity, Authority, Evidence and Receipts

How identity, authorisation, evidence and receipts work together without pretending to solve every trust problem.

Project statusLive public MVPPVTY tokenLive on Solana
Recorded 16 August 2026:This is a historical educational recording, not a live status report. Use the current evidence links below for product, supply and market status. Future utility described in the recording is not a deployed benefit.

Five-minute narrated explainer · original publication: 16 August 2026. Page and evidence links reviewed 27 September 2026.

Watch on the official YouTube channel ↗

Recheck the current evidence

Original narration transcript

The transcript preserves the historical recording. It is not rewritten to imply that later developments were already live when the video was recorded.

An autonomous agent needs more than a name or wallet. A practical trust stack separates several questions. Who or what is acting? What is it authorised to do? Which policy applies? What evidence shows what happened? Can another party recheck the result? PactVerity focuses on the evidence and receipt layer while recognising that identity and authorisation must connect to it.

Identity can link an agent to an organisation, operator, key or credential. It may also describe the software version or deployment environment. Identity establishes attribution, but it does not show that an action was allowed or completed correctly. A well-designed receipt references the relevant identity context without exposing unnecessary personal or secret information.

Authorisation states what an agent may do: which API it can call, what value it can spend, which data it can access and when it must ask for human approval. Capability-based permissions and policy-bound wallets are useful controls. Yet authorisation is an ex-ante rule. After execution, a reviewer still needs evidence of the action taken within that boundary.

Evidence may include timestamps, response measurements, signed events, digests or other bounded observations. Collection should be privacy-conscious and proportionate. An evidence system should not copy entire datasets when a minimal proof is enough. It should also state what it did not observe. PactVerity receipts are intended to preserve the supplied evidence context, not claim invisible access to the outside world.

The same evidence can produce different outcomes under different policies. A five-second response may pass one agreement and fail another. Recording the policy version prevents the evaluation rule from silently changing after the event. Deterministic checks give reviewers a visible path from term to evidence to result, even when the policy itself remains open to criticism.

The receipt combines the relevant identity reference, terms, evidence summary, policy version, checks, outcome and timestamp. A digest helps expose later alteration. Another compatible system can verify the receipt's internal consistency. This creates a common technical artifact for customers, providers, agents and auditors without forcing every party to rely on one private dashboard.

Some decisions require escalation, appeal or interpretation. A receipt can make the facts and policy path clearer, but it cannot settle every ethical, legal or subjective dispute. High-stakes medical, legal or physical-world outcomes need stronger safeguards than a deterministic digital receipt. Good governance uses receipts as evidence within a broader decision process.

Map one workflow across identity, authorisation, evidence, policy, outcome and review. Look for gaps where a later investigator would have to trust a screenshot or guess which rule applied. PactVerity's public Proof Engine and Receipt API beta provide a test surface for that exercise. Visit pactverity dot com to inspect the current implementation and its documented limitations.

Continue the series