Questions? Answers.
Can't find what you're looking for? Reach out.
Why switch from BigQuery Studio?
BigQuery Studio is built to manage your cloud. Querylab.io is built for writing SQL: completions from your real schema, errors caught before you run, and a cost estimate on every query.
Can I try it before committing?
There is nothing to commit to. Querylab.io is free, including commercial use. No credit card, no trial timer, no seat count. The Playground lets you explore without even connecting BigQuery.
What can Querylab.io see?
Not your query results. Queries run from your browser or the desktop app straight to BigQuery with your own Google credentials, and results come straight back. Your SQL and tabs stay on your device unless you turn on cloud sync. On the web, your Google token is stored encrypted on our servers so your session stays signed in. On the desktop, it stays in your Mac's Keychain or the Windows Credential Manager. Prism AI sends SQL and table names to the AI provider, and only when you use it.
My company blocks third-party Google apps. Can I still use it?
Yes. The desktop app can sign in with your existing gcloud credentials (gcloud auth application-default login). There is no Querylab.io grant for your admin to approve, and the credentials stay on your machine.
Can my own AI tool query BigQuery through Querylab.io?
Yes. The desktop app runs a local MCP server, and Claude Code, Claude Desktop, Codex, Gemini CLI, Cursor and VS Code each connect to it with one button. Every call they make is held to the limits you set: a maximum GiB per query, a daily limit for AI tools, per-dataset rules, and writes that wait for your approval. The MCP activity tab lists every call with what it billed and what it cost.
What makes Prism AI better than Cursor with MCP?
You can have both: the desktop app's MCP server lets Cursor or Claude Code query BigQuery through your cost limits. What Prism AI adds is the rest of the work in one window. It searches your real schema before it writes SQL, dry-runs what it writes, puts the price on the Run button, and leaves the result in a grid with history and exports.
What else does Querylab.io do?
Column lineage, notebooks, query history across every tool, data import and compare, dbt and Dataform support, GA4 exports, time travel, and a local BigQuery API on the desktop. The All features tab lists each one with a link to its docs.
How does it reduce BigQuery costs?
Tablesample mode runs test queries on a sample of each table, at a fraction of the cost. BigQuery can't sample views, and a table up to about 168 MiB is one block, so those are read in full. Budget limits block runaway queries. Per-CTE breakdown pinpoints what's expensive. Auto routing sends each query to on-demand or to your slots, whichever its past runs say is cheaper.
How long does it take to get started?
Two minutes. Sign in with Google, authorize BigQuery, start writing. No installation, no config, no onboarding calls. Schema loads automatically.
Is there a desktop app?
Yes, for macOS (Apple Silicon), and in beta for Windows from the Microsoft Store. Credentials stay local in your OS keychain, hotkeys are native, Prism AI drives the agent already on your machine, and the local MCP server lets that agent query BigQuery inside your limits.
Where can I learn BigQuery SQL?
On learn.querylab.io, for free. It has 76 lessons and 251 challenges, from SELECT and joins to window functions, cost control, scripting and GA4 analytics. You write real BigQuery SQL in your browser, and every answer is checked. There is no account, no Google Cloud project and no card.
