JEV USE CASE
Browser Automation.
Turn the current page into a closed set of possible actions and ask Jev which action best advances the goal.
THE DECISION CONTRACT
STATE
Goal, visible text, current URL and a numbered list of clickable elements.
QUESTION
Which available action should happen next?
CHOICES
click_3 click_7 scroll stop ask_user
TYPESCRIPT EXAMPLE
Ask for one bounded answer
This Vercel AI SDK-shaped example is illustrative. Check the current provider documentation before shipping.
import { experimental_evaluate as evaluate } from 'ai';
const result = await evaluate({
model: 'typesafe-ai/jev',
state: "Goal, visible text, current URL and a numbered list of clickable elements.",
questions: {
decision: {
type: 'choice',
instructions: "Which available action should happen next?",
options: ["click_3","click_7","scroll","stop","ask_user"]
}
}
});
// Validate permissions and thresholds before acting.
console.log(result.answers.decision);SAFE IMPLEMENTATION
Keep the boundary clear
Guardrail
Jev does not see or click by itself in this pattern. Code builds the action set, enforces domains and performs the action.
- Log the input, answer distribution, chosen threshold and resulting route.
- Test on labelled examples from your own production distribution.
- Send low-confidence, high-impact or out-of-distribution cases to a safer fallback.
- Re-evaluate thresholds when the model or workflow changes.
PRIMARY SOURCES