JevMade Sign in
← Back to experiments

Browser & computer use

Lizard

A browser agent with no LLM in the loop: code turns the page into a closed set of actions and Jev decides the next one, its target, and whether the task is done.

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

Only you can see your notes.

Source screenshot of Lizard
SOURCE SCREENSHOTFull screenshot ↗

What it does

Prices and dates go through deterministic constraints, and an answer is located on the page rather than composed. Reusing the connection cut a decision from 1,066 to 349 milliseconds.

How you can use it

Your developer can adapt these Python scripts to automate website tasks. For example, it can search a store for a specific product. The tool lists the visible buttons and links. It then asks an outside AI service called TypeSafe to pick the next step. This requires a TypeSafe access key.

The program compares prices and dates directly instead of asking the AI. It stops automatically if a website shows a security check. The tool only picks from real items on the screen. It cannot try to click a button that does not exist.

Maker-reported (not independently measured by JevMade): $0.000092–$0.000194 per perception call · a 30-step task costs roughly $0.0034 · 1 question 368 ms, 5 questions 369 ms, 15 questions 356 ms

Primitives
choice, noul, score
Platform
Python
Added
Project created

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.