Build an approved brand fact base
Turn approved product, pricing, and specification information into atomic facts with sources.
When to use this workflow
“AI confuses our product specifications; prepare the facts it should be checked against.”
Before fetching
Confirm what is sold, what AI gets wrong, which figures may be stated, and the product lines in scope. Require authoritative URLs or supplied reference files.
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. Resolve the brand and products
Read the brand and observed products to identify which product each fact describes. Do not assume the observed set is complete; distinguish additional products from the supplied catalog.
orc brands get YOUR_BRAND_ID --json
orc products list --brand-id YOUR_BRAND_ID --json2. Record one sourced fact per statement
Have the agent read supplied product, pricing, and specification pages. Record each fact, product, source quote, URL, capture date, and applicable period. Cover every product line; never substitute the nearest fact from another product.
3. Use only approved facts for verification
Mark undocumented prices or capabilities as unverified and deliver the facts in a reviewable file. Keep conflicting or differently dated information separate until resolved.
Deliverable
Deliver facts by product, source quotes and URLs, applicable periods, approval states, and open questions.
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.