Does iDBQuery sample large tables or read all the data?

A frequent worry is that an AI tool quietly answers from a small sample, giving numbers that are close but wrong. iDBQuery is built the opposite way.

• Aggregation pushed to the source. Instead of pulling millions of rows into memory, iDBQuery writes SQL that sums, counts, and groups inside your database or warehouse. Only the compact result travels back, so "total sales last year" is computed over every qualifying row.
• Query-in-place. Because it queries your live data where it lives rather than force-copying it, there's no stale extract to answer from.
• Sampling only for orientation. When you're exploring — "show me what this table looks like" — it may display a handful of representative rows. That's a preview, clearly, not the basis of a calculated figure.

This keeps answers both accurate and fast on large datasets: the database is very good at aggregating billions of rows, so iDBQuery lets it. And since every figure is cited back to its exact source rows and the SQL is visible, you can confirm the answer used the whole population, not a shortcut.