Measures What Actually Runs
Reads the statistics PostgreSQL, MySQL and CockroachDB already keep for every query, and Trino's query events. Each execution counts once. No proxy in front of your traffic, no SDK in your application.
Features
Everything you need to make your database faster.
QueryInsights handles the full SQL performance loop: it measures what your database actually spends time on, proves which schema changes help the whole workload, and hands you changes you can ship, without ever getting in the way.
Sees the Patterns, Not the Noise
Thousands of queries collapse into a handful of patterns by replacing literals with placeholders. You see what's truly heavy — not duplicates dressed up as different queries.
Rewrites Slow Queries
Every slow query pattern gets a focused AI query-optimization review grounded in your real table definitions — predicate ordering, join shape, engine-specific gotchas, written so an engineer can ship it.
Proves Index Changes on a Twin
Candidate indexes are tested on an empty copy of your schema with production's statistics. Your database's own optimizer costs the whole top-N workload, and any change that slows a query is rejected.
Ranks by Real Database Time
Patterns are ranked by total execution time, calls and rows read from your database's own counters. Traffic you can't change, like a Spark job, can be ignored or kept for schema advice only.
Finds Unused and Duplicate Indexes
Indexes that no query reads, and indexes another index already covers, are found straight from the catalog with no AI involved. Each comes with the DROP statement and a caution about when usage counters reset.
Two kinds of advice
Per-query rewrites, plus workload-wide schema thinking.
These two stay deliberately non-overlapping so the advice you get is always actionable, never contradictory.
Make this specific query better.
Each unique pattern gets focused AI review grounded in your real table definitions. You get a rewritten query, not generic pointers — predicate ordering, join shape, projection scope, and engine-specific nuance, all in one place.
Make the whole workload faster.
Index changes are searched and costed on a workload twin across your top query patterns at once. You see a set only when it lowers total cost and no query gets slower, with each query's plan before and after.
Early access
Want to see what's slowing your database?
QueryInsights is in early access. Tell us a little about your workload and we'll be in touch.
Or email us at bhargav.akhani@queryinsights.dev