Daily work
Activity and the run log
Activity is one feed for everything that needs a person, across every brand you can see. Open it from the workspace sidebar, or a brand’s Activity for that brand alone. The bell at the top right of every page opens it too.
The feed
Events are grouped into Today, Yesterday, This week and Earlier. What lands here:
- Real changes: a brand moved outside its range.
- Runs: a weekly run finished with some prompts left out, because too few answers came back, a Check now finished and its answers are ready to read, or someone flagged a wrong mention.
- Budgets: a run paused, or started using credits after the brand’s allowance ran out.
- Reports: a monthly report was drafted and is ready for notes, or someone outside the workspace opened a shared report (at most once a day per report; your team’s own views don’t count).
- Team: someone accepted an invite and joined the workspace.
Filter with All or Unread, by Type, and in the workspace by Brand. Click an event to go to what it’s about.
Read and unread
An event counts as unread until you open it, and the sidebar shows how many are waiting. Read state is saved for you. Events older than three days count as read. Mark all read clears the rest.
The run log
In a brand’s Activity, choose Run log. Each run shows:
| Column | What it shows |
|---|---|
| Started | When the run opened |
| Trigger | Scheduled (the weekly run), Onboarding (the first check) or On-demand (Check now) |
| Status | Running, Complete, or Partial when answers are missing or the run couldn’t start |
| Answers | How many completed, out of how many were expected |
| Retries | How many answers needed another try |
| Detail | Repeats, credits used, or why a run paused |
Each answer costs one unit. A prompt and engine pair with fewer than 3 completed repeats is left out of that week’s numbers.
Why runs fail or pause
Engines sometimes time out or refuse a request. Ocita tries each failed answer again, up to 3 times. A run that ends with answers missing is Partial, and the Detail column says how many prompt and engine pairs were left out.
A run that couldn’t start shows the reason here and in the feed: the trial ended, a payment failed, the brand is archived, runs are paused, or the brand reached its monthly answers. See Troubleshooting.