My AI Agent Did Not Do What I Asked: Proof

Compare said and done

Put the original request, agent summary, receipt, and current system state side by side.

Find the boundary

Identify whether the issue was a missing call, invalid input, provider failure, blocked permission, or unconfirmed outcome.

Choose the next action

Retry only safe operations, ask for approval when needed, and preserve the failed record for review. Link the related said versus done guide.

Frequently asked questions

Who authored the AER-1 draft?

AI agent did not do what I asked proof

When an AI agent did not do what you asked, compare the request with the recorded tool calls, observed results, and final state. A receipt can show whether the call ran and what it observed, while a read-back can show whether the intended outside change exists.

What can a verifiable receipt prove?

It can show what the execution layer recorded and what its integrity checks verify. It cannot prove an unobserved outside outcome.

How can another reviewer check the record?

Open the public receipt, compare the tool, time, result, and status, then call the verifier. Keep outside confirmations separate from the execution record.

What this question can establish

The practical question behind My AI Agent Did Not Do What I Asked: Proof is whether the record supports the claim being reviewed. A user believes an agent reported success but the requested change is missing or different in the system that should own it. Separate the request, the execution, the system's observed output, and confirmation from any outside system. These are different evidence boundaries, and a reviewer should not treat them as interchangeable.

Retry only operations known to be safe, request approval when needed, and preserve the failed record so the next attempt has context. A useful conclusion names the evidence that was checked and the conditions that remain unresolved. This keeps an execution record useful without stretching it into a broader assurance claim.

Evidence to retain for this workflow

Start with the smallest set of fields that lets a second reviewer identify the event and repeat the relevant check. For this topic, Compare the original request, agent summary, recorded calls, returned results, failure states, and current owner-system state.

• Task boundary: record the requested state and what would count as acceptance.
• Execution identity: preserve the receipt ID, tool name, caller scope, time, and status.
• Committed result: keep canonical bytes or the published representation needed to recompute the output hash.
• Provenance: distinguish an action the system ran from one it observed or merely received as a report.
• External confirmation: capture a real callback, transaction reference, read-back, or provider record when the task depends on outside state.
• Review context: note who checked the record, which verifier was used, and any limitation that affected the conclusion.

A repeatable review sequence

Use the same order for routine checks and incident review. It prevents a later summary from replacing the evidence the task actually requires.

1. State the requirement. Write the expected outcome and the exact source that can confirm it. A user believes an agent reported success but the requested change is missing or different in the system that should own it.
2. Locate the execution. Match the public receipt identifier to the requested event, then confirm the tool, scope, timestamp, and status.
3. Check integrity. Decode the committed bytes as specified by the receipt format, recompute fingerprint, and compare the complete digest rather than a shortened preview.
4. Check the owning system. When the claim involves an outside change, query that system or inspect its genuine callback. A receipt alone cannot supply a missing confirmation.
5. Write a bounded finding. Record what ran, what was observed, whether the bytes matched, and what remains unknown. Retry only operations known to be safe, request approval when needed, and preserve the failed record so the next attempt has context.

Failure states and misleading success signals

Preserve failed, blocked, pending, retried, and partially completed states as distinct outcomes. Classify the gap as no call, invalid input, blocked permission, provider failure, partial result, stale read, or unconfirmed outside effect before retrying. A successful transport response can still contain a tool error, and a valid integrity check can still protect an incorrect or incomplete result.

When evidence conflicts, keep both records and identify which system owns each fact. Do not overwrite an earlier receipt or silently promote a retry into proof that the original step succeeded. Explain whether the conflict is a timing issue, a stale read, a missing callback, a policy boundary, or a different task interpretation.

Privacy, retention, and safe sharing

Reviewability does not require publishing every input. Keep account identifiers and private task content out of public examples. Use a sanitized receipt and a controlled read-back for diagnosis. Prefer stable references and cryptographic commitments where they are enough for the reviewer, and keep the underlying sensitive record with the system that controls it.

Before sharing, check the public page, result preview, metadata, and linked evidence for secrets or personal information. Set retention and access according to the review purpose. If a record must be corrected, publish a new reference and preserve the original history rather than rewriting a record that another reviewer may already have captured.

Example of a careful conclusion

This kind of note gives the next person a useful handoff: what was checked, what the check means, and where the evidence stops. It is more informative than a bare success label and avoids presenting an unobserved event as a verified fact.

Common review questions

What does the record establish?

It establishes what the execution system recorded and whether the exposed canonical bytes match the output commitment. For my ai agent did not do what i asked: proof, it does not by itself establish correctness or a result outside that system.

What if a provider or callback is unavailable?

Keep the state pending or unavailable, retain the error and time, and avoid claiming completion. In this workflow, Classify the gap as no call, invalid input, blocked permission, provider failure, partial result, stale read, or unconfirmed outside effect before retrying.

What should a reviewer inspect first?

Compare the original request, agent summary, recorded calls, returned results, failure states, and current owner-system state. Then compare the record with the task requirement and ask the system that owns any claimed external state for a separate read-back.

What is safe to share?

Share the public receipt URL, identifier, status, verifier result, and a short sanitized task description. Keep account identifiers and private task content out of public examples. Use a sanitized receipt and a controlled read-back for diagnosis.

Further reading

• Trust and verification
• Reviewing agent work
• Execution evidence boundaries
• Task completion checks
• Agent handoffs
• receipt definition