JevMade Sign in
← Back to experiments

Agent tooling

jev-workbench

A local workbench for building and publishing versioned Jev judgment functions.

Bookmark: jev-workbench Keep this in your collection.
Leave a noteWhat would you try with this? : jev-workbench

Only you can see your notes.

Source screenshot of jev-workbench
SOURCE SCREENSHOTFull screenshot ↗

What it does

Published functions can be called from backends over HTTP or from coding agents over MCP, while the vendor key remains on the user's machine.

How you can use it

A developer can install this tool on your computer. To make live decisions, you need an access key from TypeSafe, the company running the AI service. The tool keeps this key safe on your computer instead of sharing it with outside callers.

Before building, gather a few sample customer requests and decide which department should handle each one. You can open the tool in your browser, set up questions to sort those examples, and test them. Remember that the AI only estimates answers and can make mistakes.

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.