Jev AI Tools

Jev AI Tools is an independent, hands-on lab for exploring TypeSafe AI’s Jev evaluation model. It helps developers see how a model can return a defined choice, a score on a rubric, or a probability for a yes-or-no question. The site shows the model’s answer separately from the application rules that decide what happens next. It is not affiliated with TypeSafe AI.

The homepage starts with a customer message example. Visitors can inspect a prepared sample without an account, change the message, and review how the returned signals feed a routing policy. Six interactive experiments cover startup idea assumptions, draft tone and clarity, an agent’s next allowed action, task routing, context selection, and claim-versus-evidence checks. Each experiment explains its boundaries and keeps the model result visible so developers can judge whether a policy should act, pause, or ask for human review.

Jev answers structured questions about text or data supplied by the developer. A choice question picks from named options and exposes probabilities across the candidates. A score question rates an input against an ordered rubric. A boolean question estimates the probability of a statement rather than reducing it to a flat yes or no. The playground illustrates how these values can be consumed by TypeScript code without parsing a paragraph of generated prose.

The site also includes a cost calculator, an MCP configuration generator, and an API request builder. Its guides explain Jev’s answer types, pricing, API usage, MCP setup, and practical limitations. An ecosystem directory links to public SDKs, tools, and related projects. Synthetic examples are available even when live model runs are unavailable, and live usage is subject to the site’s service limits.