Run an audit
Starts the collection pipeline for one client against the engines you pick. No run credit is consumed. The run is detached, so you can close this page and it keeps going. Follow it, and read its report afterward, on Audit Runs.
Client
Active clients only. Pick one to see its cohort size and which engines its tenant has enabled.
Engines
Default is every engine this tenant has enabled. Engines the tenant has switched off cannot be selected here — turn them on in the tenant's add-ons first. Copilot runs through SerpApi and needs SERPAPI_API_KEY on the pipeline box. Grok and AI Mode are absent because the runner has no collection path for them yet.
Passes per prompt
Three is the framework default. AI answers are probabilistic, so a single pass cannot separate a brand that is genuinely absent from one answer that happened to omit it. Use one pass for a cheap check, never for a baseline.
Before you start
Recurring schedule
Separate from the run above. This sets the cadence the hourly scheduler reads, and it takes effect whether or not you start a run now. Paused is the off switch. Scheduled runs use the pipeline defaults, not the engines and passes selected above.
Jev answer readings
The scheduled job that reads every completed weekly cycle with Jev: the second reading of each answer, consistency, stale facts and objections. Until it has run, Prompt View, the standing-still panel and Brand Answers say "not read yet". This runs one tick now, for every website, over the job's lookback. It spends on TypeSafe and answers with the reason when a server switch keeps it from reading. To see whether Jev read one audit, or to read just that one, open it on Audit Runs.