Fair-use policy

Chat is unlimited on paid plans for people doing their own work. Fair use applies to each person separately and sits well above normal use, even heavy daily use; it exists to stop automated or shared-account abuse. If someone reaches the fair-use limit, chat pauses for that person and we are notified. We get back to you, usually within a day, and files, dashboards, and AI credits keep working in the meantime. AI credits are only used for bulk work: 1 credit = 1 PDF page or 1 enriched row, and a deep research report is 20. Chat never uses credits. Pro includes 3,000 AI credits a month and Team 10,000, shared across the workspace, plus 1,000 per extra person, with top-ups at $10 per 1,000. Max and Business plans from before November 2026 keep at least the allowance they had, now counted in AI credits, which go further: a PDF page is 1 credit instead of 10 AI Actions, and a research report 20 instead of 50. Starter and Embed keep their AI Actions.

We count rows returned by source systems before filters, joins, or other transformations, once per successful logical run. If the final output has more rows, the larger number counts. Automatic retries keep the same logical-run identity and never count twice. A successful result with zero rows counts as a Free test.

Before a run is queued, Better Analyst reserves an estimate based on the flow’s previous successful run or its bound source artifact. When neither estimate exists, it reserves 5,000 rows and corrects that to actual rows during the run. Success replaces the estimate with actual source and output rows; terminal failure or cancellation releases it. Usage continuously falls out of the window 30 days after it was recorded—there is no fixed calendar reset day.

Workspace owners see warnings at 70% and 90%. At 100%, new manual runs are rejected with the next relief time. Scheduled occurrences are skipped while their next scheduled time continues to advance, then resume automatically when rolling usage falls below the limit. We do not bill automatic overages.

The effective limit is always the most restrictive of the Better Analyst plan, connector, source API, and destination. If a run exceeds a limit, it fails before the destination write. We never truncate a result or partially replace a destination to fit a quota. Provider latency, API quotas, spreadsheet cell limits, and database constraints can impose lower limits.