jev
A decision layer for your AIExplore Playground

Jev AI Decision Models for Your App

Jevai (jevai.pro) is an independent Playground and guide to TypeSafe's Jev decision model. Try your own context and choices, explore editable examples, and learn how to route requests, select tools and classify text.

Try an example

Choose an action from interface text, rather than generate a browser command. Try replacing the entire goal with a request to update a saved card.

Source: X · @milindlabs. Inspired by the author's UI-element selection demo. This is our text-only scenario; it does not control a browser or reproduce the reported timing.

Live API

Which visible control should the agent select next to move toward the stated goal? Choose No suitable control if none helps. Select only; do not perform an action.

View receiptsPayment methodsManage planContact salesNo suitable control
Customize question and choices

Edit any input to explore a different decision. No request is sent until you run it.

Your next decision

Not run yet

Choose an example or write your own input, then run it to see the ranked choices.

Results are not stored by this site. Your submitted text is processed by the API provider.

Built around the next step

Route requests, classify text, and review uncertain decisions.

Give every route, label, and tool call a clear decision boundary.

Choose the route. Then let the model work.

Put a focused decision before your next model call. Define the routes your application supports and give every request a deliberate starting point.

Explore model routing
Illustration of one request selecting the reasoning route from fast, reasoning, and review options.
One request. A finite set of routes.

Turn incoming context into an actionable label.

Bring the situation, ask one specific question, and name the valid outcomes. A support ticket becomes a team your application can send it to.

See the support recipe
Illustration of incoming support messages classified into Billing, Shipping, and Technical teams, with a review route.
Context → question → choices.

Keep the uncertainty where you can see it.

Inspect the ranked choices, then set your own policy. Confident results can move forward; uncertain ones can stay in a review queue.

Test a decision
Illustration of ranked choices with a highlighted leading result and a separate human review route.
Inspect the choice before the action.
Who uses Jev

For teams building decisions into software.

Use Jev when a small set of routes needs to be inspected, measured, and acted on.

Builders

AI product teams

Turn routing and classification rules into an explicit request contract.

Support

Operations teams

Separate routine tickets from cases that need a human or deeper reasoning.

Agents

Agent builders

Choose one tool or route from a controlled list and keep the fallback visible.

Research

Model evaluators

Compare fit, confidence, latency, and cost with a repeatable shape.

How it works

Context, question, choices, then a result.

Give the model a situation, a question and a set of choices. Get a decision your code can use.

  1. Illustration of a delayed parcel and a customer message asking where the package is.

    Start with context

    Describe what happened in the same words your application receives.

  2. Illustration of one focused support-routing question: Which team?

    Ask one question

    State the decision clearly, so the model knows what to resolve.

  3. Illustration of four allowed team choices: Billing, Refund, Shipping, and Technical.

    Define the choices

    List the valid routes that your application can act on.

  4. Illustrative Shipping choice with ranked result bars and a handoff to application code; not actual API output.

    Use the decision

    Inspect the leading choice and score before taking the next action.

Model landscape

See the decision model landscape at a glance.

Find the model, see who builds it, and explore where it fits in your workflow.

Convai Innovations

Laya

Decision model
Typical use
Local decision tasks
Output
Profile under review
Profile under review
Independent project / wfzyx

Von

Decision model
Typical use
Decision experiments
Output
Profile under review
Profile under review
Multiple providers

LLM

Language model
Typical use
Open-ended reasoning
Output
Generated text
Compare with JevGeneral model category
Start with a scenario

Find a decision you already need to make.

Illustration of one request selecting the reasoning route from fast, reasoning, and review options.Use case

Choose a model or human review

Use an explicit policy to choose a fast model, a reasoning model or human review for the next task.

ExploreIllustration of incoming support messages classified into Billing, Shipping, and Technical teams, with a review route.Recipe

A home for every ticket

Classify incoming support requests with an explicit human fallback.

ExploreIllustration of ranked choices with a highlighted leading result and a separate human review route.Model

Meet the decision model

Understand the input contract, output format, and task fit.

Explore
Before you build

What to know before using Jev.

A few things to know about the model and the way to use it.

What is Jev AI, and what can I do on this site?

Jevai (jevai.pro) is an independent site for trying and learning about Jev, TypeSafe's decision model. Use the Playground and editable examples to choose a route or label from supplied options. Try an input, inspect the returned choice and probabilities, then follow a workflow guide or recipe to plan an integration. TypeSafe provides the model; Jevai provides this site's interface and guides.

When would I use Jev instead of a generative LLM?

Use the Choice workflow here when your application already knows the allowed outputs, such as teams, tools or model routes. A generative model remains useful when the task needs a written answer. The two can work together: a decision selects a handler, and your code runs the next step.

Can I test my own input and choices?

Yes. Edit the context above and open Customize question and choices, or use the full Playground for single and batch text tasks. A request runs when you submit after security verification. Examples select labels; they do not execute tools, send replies or change your accounts.

Does a confident result mean I can automate the action?

Confidence describes the returned distribution, not a guarantee that the decision is correct. Evaluate representative inputs, include a review or no-match path, and keep permissions and business rules in your application. Start with the Playground, then follow the support-routing recipe and official TypeSafe documentation.

Make your next AI decision inspectable.

Try the playground with your own context and choices, then bring the same decision into your project.

Try PlaygroundTry support routing