Browser Use + Jev
Gregor Zunic's flight-search demo with a dynamic DOM action space.
A FIELD GUIDE TO JEV / VOL. 01
The best of Jev. Real projects, practical guides and ideas from across the internet.
Gregor Zunic's flight-search demo with a dynamic DOM action space.
01 / A small decision engine. A whole new way to build.
1,195 resources / The collection
SOURCE-LINKED, ALWAYS.AI-assisted summaries and translations. Check original sources for context and performance claims.
Real-time virtual try-on experiment using Jev to select and change outfits based on user input and transcript analysis.
Jev controls Super Mario Bros. using fast inference and structured outputs, demonstrating potential for real-time applications.
Browser Use's ultrafast agent: Jev picks the operation and DOM element in one request; a small LLM only writes text when typing is needed.
Experiment combining JSON rendering with Jev to create instant generative user interfaces for rapid component and design system rendering.
Shortest path from an API key to a typed decision.
Claude Code plugin that replaces the compaction summary with Jev decisions: every tool call and result is scored in one fast request, stale ones are dropped or truncated, everything kept stays verbatim.
Demo showing Jev selecting the most suitable AI models for video and image generation tasks on Higgsfield.
Classifier that assesses prompt difficulty in real-time, enabling fast mode for simple prompts to enhance user interaction.
Simulation where MOSS picks up litter based on instructions, aiming to improve street cleanliness through automated decision-making.
The case for machine-native intelligence built for software, not conversation.
Launch post: architecture, RLCD training, pricing, Doom and Wikiracing demos, FAQ.
Request and response contract for POST /v1/systemone.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Official sync and async Python client. pip install typesafe-sdk.
Paste a state, add questions, and see typed answers in the browser.
Choice, Score, and Noul: the three question types and what they return.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Reproducible recipes: parallel questions, reranking, guardrails, citation checks, extraction, hierarchical classification.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Known failure modes of the current public model, documented by TypeSafe.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Product and research updates.
Introduction, primitives, patterns, cookbooks, HTTP API, and SDK references.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Official TypeSafe reference: Introduction.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Official TypeSafe server. Builder demos live in the Show and Tell channel.
Speculative fan-out, confidence-gated routing, composite scoring, and intent routing.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.Published eval methodology and per-model results for System One workflows.
Official interactive demo of speculative fan-out: many questions in one call, code keeps the relevant answers.
const answer = await client.systemOne({
state, questions
});
// intelligence, in your code.