Runs and usage

What a run is

A run is one round of checks. AnswrForge asks each selected prompt on each selected provider, saves the answers, then analyzes them.

Runs start manually or on your schedule, and the Runs list marks each one Manual or Scheduled. Only one run can be active in a project at a time.

Start a check manually

1. 1Open Runs and click Run check at the top of the page.
2. 2On Confirm visibility check, review Prompts, Providers, Samples each and the expected answers.
3. 3Click Confirm run.

Run check includes every enabled prompt on each of its enabled providers, even those set to Manual only. To check a single prompt, select it in Setup under Prompts and click Run now.

If settings change while you review, AnswrForge updates the preview and asks you to confirm again. Run check is unavailable while another check is active.

Follow run progress

The Run progress page shows the run's status: pending, running, completed, completed with errors, failed or cancelled. It counts answers collected and analyzed.

Progress by provider breaks this down for each AI assistant. While a check runs, a banner shows A check is active. with a View active run link.

Handle failures

If checks fail, Collection failures or Analysis failures explain why. Click Inspect failed checks to see them.

Click Retry failures to try again; when only analysis failed, saved answers are reused. Some failures retry automatically, and the page shows when.

Cancel outstanding work stops pending work. Requests already running may finish, and collected answers stay saved.

Review past runs

Open Runs in the sidebar. Run history lists each run's Status, Date, Run type and progress. You can filter it by Acceptance period.

Expand Scheduled runs to see recent automatic runs, including skipped ones, and the next scheduled run.

Understand your usage

AnswrForge measures work in checks. One check is one prompt asked once on one provider in one country.

• Before each run, the confirmation page shows how many answers to expect and how many analyses follow.
• Samples per prompt/provider in Setup under Sampling (1, 3, 5 or 10) multiplies your answers. Saved setup workload shows planned checks per run.
• Several countries add separate checks where supported.
• Schedule shows a monthly estimate of future checks. It estimates future work, not usage already incurred.
• Each project has a daily request allowance that resets each day (UTC). If it runs out, AnswrForge tells you; wait for the reset before starting or retrying checks.