What it does
It supports TypeScript and Python, with policy checks for capabilities, context, and budget.
Agent tooling
tiershift
A model router that uses Jev and a YAML policy to choose the least expensive LLM suited to each request.
Bookmark: tiershift
Keep this in your collection.
Leave a noteWhat would you try with this?
: tiershift
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
If you manage an automated customer support chat, you can borrow this routing approach to lower your operating costs. Instead of directing every routine greeting to a premium, expensive AI model, the program sends straightforward messages to a cheaper option and reserves advanced models for difficult questions.
A developer can install the tool or adapt its rule file to review incoming requests. You will need an access key from a TypeSafe account to evaluate question difficulty. Keep in mind that this setup was tested on single questions, so longer tasks with multiple steps still need top-tier AI.