Can Jarvis AI Assistant Explain SQL Queries?

A SQL query on screen is something Jarvis (getjarvis.eu), a screen-aware desktop AI assistant, reads back in plain English. Press Cmd+/ on macOS or Ctrl+/ on Windows over your editor, database client, or dashboard and ask what the query returns. It walks through the joins, filters, grouping, and window functions, then can flag a performance problem or suggest a rewrite. It can spot performance issues — a missing index hint, a needless subquery, a Cartesian join — and suggest a rewrite. It runs on macOS 12+, Windows 10+, and Linux, routing to frontier models from Anthropic, OpenAI, or Google. Data is AES-256-GCM encrypted, GDPR-aligned, and never used for training. A free plan (40 requests/week); paid plans from $16/month.

A 60-line query with five joins, a CTE, and a window function is hard to hold in your head. Jarvis reads it as displayed and explains the shape: which tables join on what keys, what the WHERE actually filters, how the GROUP BY rolls up rows, and what each window function computes. It then describes the resulting columns and granularity — one row per user, or per user per day? Getting the grain right is where most SQL confusion lives, and that's exactly what Jarvis pins down.

Correct SQL can still be slow. Jarvis flags common culprits visible in the text: a join missing a condition (accidental Cartesian product), a function wrapping an indexed column (defeating the index), SELECT * pulling unused columns, or a correlated subquery that could be a join. It suggests rewrites and explains the tradeoff. It can't see your query planner's EXPLAIN output unless it's on screen — but if it is, Jarvis reads that too and interprets the plan's bottleneck, like a sequential scan on a large table.

Code & tickets