Code mode does the heavy lifting.

For big jobs the agent writes JavaScript instead of calling tools one by one. The code joins, aggregates, and exports across your datasources in a sandbox, and the result lands as a file in your workspace.

Book a demo

One run instead of twenty tool calls.

The agent writes sandboxed JavaScript, runs it, and drops the result as a file. You can watch the code and cancel it.

const orders = await query("shop", { from: "orders", … });
const subs = await query("billing", { from: "subscriptions", … });
const ltv = joinOnCustomer(orders, subs);
await writeFile("ltv.csv", toCsv(ltv));

ltv.csv · 48,102 rows · 1 run

Rows stay out of the chat.

The code fetches and crunches the data inside the sandbox. The model sees the summary and the file name, not a million rows of context.

1,204,882 rows scanned in the sandbox

48,102 rows written ltv.csv

4 lines reached the model the summary

Code is not a back door.

Every query the code runs goes through the same access policies as the ordinary tools. A column you hid from the AI stays hidden inside code mode too. Runs live in an isolated worker you can stop at any point.

await query("shop", {
  from: "customers", select: { email: true },
});

Denied · customers.email is hidden from the AI

Sandboxed worker, no filesystem data tools only

5 minute timeout guaranteed kill

100 KB chat output cap big results go to files

Cancel button on every run stops mid-execution

Let the agent do the heavy lifting.

Free for personal use, every feature included. Commercial use needs a seat at beta pricing. See pricing.

All downloads Other platforms and builds