37 reports, and 11 that tell you something new

No, and we used to say they were until we corrected it. They contain no model call — they are queries and heuristics over your own execution history. That means they consume no AI credits, they work the same on a free plan, and they are deterministic: the same data gives the same answer every time, which is what you want from something informing a release decision.

Mostly patterns across time that no single view shows: which tests change result without the code changing, which cases nobody has touched in months, which parts of the suite are redundant, and whether your testing throughput is falling. Any one of those is visible if you go looking; the point is that nobody does.

Execution and coverage reporting works from your first cycle. The derived ones need history — flakiness, velocity and predictive failure are statements about patterns, so they say little in week one and get sharper as executions accumulate.

Yes, where those results are reported into the product — through the CI integrations, the webhook, or the automation reporters. A report describes what was recorded, which is also the usual reason a coverage figure looks lower than a team expects: testing done outside the tool is invisible to it.

Both, deliberately split. Execution and coverage reporting is for the people doing the work and answers 'where are we'. Release Readiness and the risk heatmaps answer the question asked about the work — can we ship, and what is risky — which is what usually gets decided in a meeting on the strength of how confident someone sounds.

No — treat it as a conversation starter. It measures recorded activity, not judgement or difficulty, so it is useful for spotting an imbalance across a team and wrong as an appraisal input.

No. There is no custom report builder, so you get the reports that exist rather than a designer for creating your own. If your organisation mandates a particular reporting format, TestRail's configurability is a genuine advantage and worth weighing.