Skip to main content
Use /v2/responses for answer history and /v2/responses/{id} for an individual answer’s detail. Use citation and source endpoints for aggregate views across the workspace. Read-only access is sufficient for these exports.

​
Retrieve the answer history

Filter by the intended prompt, platform, date range, collection status, and analysis status. Paginate through the collection using the returned cursor.
The dates above are illustrative. Choose the reporting period your export needs. A list item includes an excerpt and metadata; retrieve the detail endpoint when you need the answer content rather than a shortened excerpt.

​
Combine evidence using stable identifiers

Keep response and prompt identifiers in your export so an observation can be traced to its question and answer. Preserve platform, collection date, sentiment, and citation information as separate fields. Use /v2/sources/domains and the citation endpoints to examine domain and page patterns. Do not treat an aggregate source count as the citation list for an individual answer.

​
Build a reliable reporting dataset

Keep missing values distinct from zero and pending analysis distinct from a neutral result. Make pagination resumable and retain the date window and filters used for each export. If a request fails, follow error handling without restarting every completed page unnecessarily.