liveBook · Manning

Post-execution verification

As explained in AI Governance

Post-execution verification checks whether an operation's actual results match its intended results.

For a write operation, success is not enough: compare the actual changes with the intended changes, including their scope. For a narrow operation, verify that exactly one row changed. At scale, observability platforms can log activity, but the comparison logic still needs to be built. section 4.7.1

Verify a single-record update

After an agent updates one record, check that exactly one row changed. Logging that the agent updated a record is not enough to verify the write. section 4.7.1

How AI Governance covers it

How deep it goes
Explains comparing changed records with intended scope, but leaves comparison logic to be built.
Examples
Uses a single-record update to illustrate checking that exactly one row changed.
What you can do after
Diagnose whether a write succeeded and whether its scope matched the intended change.
How it connects
Connects write verification with observability, reversible operations, prompt injection, and AI agents.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI Governance. The section links point to the parts of the book it draws on.