Does iDBQuery check its own answer before showing it to me?
Yes — iDBQuery validates its generated SQL, self-corrects errors by reading the database's response and retrying, and grounds every figure in an actual query rather than a guess. Each answer is cited back to source rows, so both iDBQuery's own checks and your verification confirm the number is real.
An AI that answers confidently but wrongly is worse than useless, so iDBQuery is engineered to check itself and to make checking easy for you.
- Query validation. Generated SQL is validated before it runs — unsafe or malformed statements are blocked — and if the database returns an error, iDBQuery reads it, fixes the query, and retries.
- Grounded, not invented. Numbers come from executing SQL against your real data, not from the model recalling something plausible. If the data can't support an answer, iDBQuery says so instead of fabricating.
- Built-in evidence. Every figure is cited to the exact source rows and the SQL is shown, which is the ultimate self-check: the answer arrives with its own proof.
For deeper questions, the Analyst agent goes further, cross-checking findings across segments and showing its steps. The combination — validated queries, honest failure, and row-level citations — is how iDBQuery keeps answers trustworthy without claiming to be infallible. You're never asked to take a number on faith; you can always trace it, and in the rare case it's off, the transparency lets you catch it before it reaches a decision.
Updated 2026-06-22