Audit the actual AI answers
Agree on an answer sample, then report wording, competitors, and claims with quotes and counts.
When to use this workflow
“What does AI actually say about us, and does it repeat the same misunderstanding?”
Before fetching
Before fetching, choose prompts, platform, dates, sample size, and sampling method. Fix the inspected set by answer ID rather than selecting only convenient examples.
Complete the CLI quickstart and confirm the workspace with orc workspace current --json. Replace example IDs, URLs, dates, and settings. Apply only approved mutations.
1. List the agreed answer sample
The example selects one prompt and platform. Specify the dates and record ordering and sample size. Continue pagination until the agreed sample is complete.
orc answers list --prompt-id YOUR_PROMPT_ID --platform openai --start-date 2026-09-01 --end-date 2026-09-07 --sort created_at --order desc --limit 20 --json2. Read each answer in full
Fetch every selected answer. Record descriptions, co-mentioned competitors, product claims, and citation URLs by answer ID. Report recurrence as n of N inspected answers.
orc answers get YOUR_ANSWER_ID --json3. Check claims against approved facts
Compare potentially incorrect claims with approved product and pricing facts. Retain the claim, reference fact, and both sources. A nearby citation alone does not establish the cause of a misconception.
Deliverable
Deliver answer IDs, sample conditions, verbatim quotes, recurrence counts, verification results, and unresolved claims.
On fetch failure, retain the error and request scope. Do not count empty data, incomplete pagination, or missing permissions as a completed investigation.
Next steps
If a failure remains, report it and verify the fix, including this workflow and the failed step.