JEVThe best of Jev. Real projects, practical guides and ideas from across the internet.
Updated 1 Resource
THE OFFICIAL SIGNAL 4 HIGHLIGHTS
Diogo Almeida@CompleteSkepticLAUNCH POSTDiogo Almeida introduces Jev, a structured decision model from TypeSafe. This resource provides an overview of the model's core approach, practical examples, and performance metrics as reported by the author.
TypeSafe AI@typesafeaiOFFICIALTypeSafe officially launches its research lab and introduces Jev to the developer community. This announcement features a launch film by Diogo Almeida and provides foundational context for the project's mission and future direction.
TypeSafe AI@typesafeaiOFFICIALTypeSafe has removed the waitlist for Jev, allowing public access to the platform. Developers can use the official console to integrate and test typed decision-making capabilities within their own software applications.
JEVLAB ARTTypeSafe AI@typesafeaiOFFICIALTypeSafe highlights Jev’s typed-output approach, where applications provide state and questions to receive structured probabilities, choices, or scores. The announcement also details the integration of the Venice API for developers.
JEVLAB ARTAI-assisted summaries and translations. Check original sources for context and performance claims.
Decision layer for coding agents: deterministic rules before any model call, then one Jev request, as a Claude Code PreToolUse hook, an MCP server, a loopback service and a team policy that personal overrides can tighten but not loosen. Ships the 300-call injection kit behind its own numbers.
JEVLAB ART
English
Chris TateThis experiment explores combining JSON rendering with Jev to create generative user interfaces. It demonstrates a method for the rapid, real-time rendering of UI components and design systems based on structured data inputs.
EnglishThis repository provides a benchmark comparing Jev against a strong LLM using the Who&When Pro agent-failure-attribution dataset. It demonstrates a methodology for evaluating how effectively these systems identify and attribute specific failures within agentic workflows.
JEVLAB ARTEnglishThis unofficial Java SDK provides integration for TypeSafe Jev and Vercel AI Gateway. It demonstrates how to implement these services within Spring Boot applications using WebClient for network communication.
JEVLAB ARTEnglishThis project demonstrates a Typesafe.ai System One model designed to navigate Neo4j graph databases. It utilizes a classifier to analyze neighboring relationships, providing a structured approach for Jev-based graph traversal and decision-making processes.
JEVLAB ARTEnglishThis project demonstrates a task routing system for specialized agents including research, coding, and writing. It utilizes TypeSafe Jev to manage agent workflows and task distribution across various functional domains.
JEVLAB ARTEnglishJev plays Tetris by making real-time decisions to move blocks.
JEVLAB ARTEnglishA community post discussing Jev's use in enforcing coding rules that linters cannot handle.
JEVLAB ARTEnglishroverlab is a 3D planetary rover simulation environment. It serves as a sandbox for developers to experiment with autonomous decision-making processes using TypeSafe AI frameworks within a virtual extraterrestrial terrain.
JEVLAB ARTEnglishA developer project using Jev for plain-language MCP tool calls without an LLM.
JEVLAB ARTEnglishThis video evaluates Jev across twelve distinct real-world scenarios. The creator shares observations regarding operational speed, cost efficiency, and practical utility, providing a subjective look at how the system performs in varied application environments.
EnglishTypeSafe AIThis interface defines the structure for handling classification results in TypeSafe AI. It provides a standardized way to access a selected label, its associated confidence score, and a comprehensive map of probabilities for all possible choices.
JEVLAB ARTEnglishBuild versioned judgment functions on TypeSafe's Jev once, then call the same published version from your backend over HTTP and from coding agents over MCP. The vendor key stays on your machine.
JEVLAB ARTEnglishGrep, but the pattern is a description. Filters lines by meaning with TypeSafe's Jev decision model: ~200 ms and a thousandth of a cent per line.
JEVLAB ARTEnglishTypeSafe AIThis documentation defines the ModelCard interface within the TypeSafe AI SDK. It outlines the required metadata structure, including name, description, and release date fields, used for identifying and managing available AI models.
JEVLAB ARTEnglishEvidence-backed index of real-world Jev (TypeSafe AI System One) use cases: repos, patterns, benchmarks, and measured results.
JEVLAB ARTEnglishRoute to the cheapest model in claude code for your task using jev-router.
JEVLAB ARTEnglishTypeSafe AIThis guide outlines the three core question types in TypeSafe AI: Choice, Score, and Noul. It explains the specific data structures and return values associated with each primitive to help developers structure their AI interactions effectively.
JEVLAB ARTEnglishTypeSafe AIThis pattern demonstrates a method for routing requests based on model confidence scores. It teaches how to implement conditional logic to ensure higher quality outputs by directing tasks to appropriate processing paths.
JEVLAB ARTEnglishJev DSPy Lab provides tools for recording, replaying, and benchmarking TypeSafe AI decisions within DSPy pipelines. It enables developers to measure calibration, selective risk, latency, and modeled costs using deterministic offline analysis.
JEVLAB ARTEnglishA drone navigation experiment where Jev makes real-time decisions to avoid obstacles in an asteroid field.
JEVLAB ARTEnglishAsk a yes/no question of every function in a codebase. Ranked answers in seconds, for cents. Grep whose pattern is a question, powered by TypeSafe Jev.
JEVLAB ARTEnglishHemant Kumar on his 151M ModernBERT reproduction: 0.83% adaptive calibration error, 3.0% of predictions flip when the option list is reversed, 35.6ms on CPU, with the failed use cases reported alongside the wins.
JEVLAB ARTEnglishThis project demonstrates an agentic browser runtime that utilizes TypeSafe Jev for decision-making processes. It showcases how the Jev model integrates with the Aside runtime to manage choice, scoring, and noul operations within a web environment.
JEVLAB ARTEnglishContext-pruning proxy for Claude Code and Codex: Jev judges which history is still needed, measured not claimed. POC here now, heading soon into https://github.com/compozy/compozy.
JEVLAB ARTEnglishPut a live Jev (TypeSafe) meter on any video: every sentence scored, rendered as a 16:9 edit.
JEVLAB ARTEnglishThis guide demonstrates how to integrate Jev with Pydantic AI agents to facilitate type-safe operations. It provides developers with a structured approach to ensuring data consistency and reliability when building AI-driven workflows within the Pydantic ecosystem.
JEVLAB ARTEnglish
Nailthy TangReal-time virtual try-on experiment using Jev to select and change outfits based on user input and transcript analysis.
EnglishThis project implements a Jev-style System One decision model using a Gemma 3 270M architecture. It demonstrates a method for achieving calibrated, rapid decision-making in a single forward pass without performing traditional text generation.
JEVLAB ARTEnglish
EthanJev identifies and manages marketing messages in Android notifications without content deletion.
JEVLAB ARTEnglishThis project demonstrates an automated agent playing Pokemon Red using PyBoy. The system manages game logic and arithmetic while the Jev agent selects branching paths, with battle outcomes evaluated against RAM state using Brier scoring.
JEVLAB ARTEnglishThis project demonstrates Jev playing the classic game Minesweeper. It serves as an experiment to showcase how the model manages real-time, turn-based decision-making processes within a structured grid environment.
JEVLAB ARTEnglishA fire evacuation simulation where each person makes Jev-style decisions.
JEVLAB ARTEnglishThis demonstration shows Jev interacting with the mobile strategy game Clash Royale. The footage captures the agent navigating game mechanics to complete and win a match, illustrating its capability to process real-time visual inputs and execute tactical decisions within a gaming environment.
JEVLAB ARTEnglishA MakerMods robot arm is demonstrated within the MuJoCo simulation environment, showing interaction through JSON state inputs and action selections.
JEVLAB ARTEnglishThis demonstration showcases Jev playing Flappy Apex independently. The creator claims the agent achieves high scores autonomously by utilizing Appduct and Fable frameworks to navigate the game environment without manual input.
JEVLAB ARTEnglishThis project demonstrates an agent designed for automated Pokémon shiny hunting. It utilizes autonomous game resets and visual checks to streamline the process of encountering rare variants within the game environment.
JEVLAB ARTEnglishThis terminal interface integrates OpenAI with TypeSafe Jev to provide answers. It demonstrates a workflow that generates transparent decision reports, allowing users to review the reasoning process behind each output provided by the system.
JEVLAB ARTEnglishThis experiment demonstrates the use of Jev to transform simple English input into elaborate prose in real-time. It showcases a creative writing tool designed to assist authors by dynamically enhancing their text as they type.
JEVLAB ARTEnglishJev AI is a System One model designed for software integration, providing typed, calibrated decisions like choices, scores, or probabilities. It aims to replace prose-based LLM outputs with structured data to enable direct, error-free programmatic decision-making.
JEVLAB ARTUXRay is a screen overlay tool designed to detect and highlight manipulative interface patterns. This project demonstrates how automated visual analysis can help users identify dark patterns in web design to improve digital transparency.
JEVLAB ARTEnglishThis resource demonstrates the use of Jev for real-time coaching support. It highlights how the tool assists users in identifying and correcting errors or inconsistencies during writing sessions to improve overall quality.
JEVLAB ARTEnglishThis tool demonstrates a Gomoku implementation using Jev. It utilizes local tactical analysis to reduce 225 potential moves to 40 candidates, from which the system selects an optimal play based on tiered strategic options.
JEVLAB ARTEnglishA community reference exploring how Jev perceives toxic content, with limited detailed analysis.
JEVLAB ARTEnglishThis project provides an open-source router that integrates Jev with LiteLLM to dynamically select appropriate language models. It demonstrates a method for implementing type-safe model routing within AI agent workflows.
JEVLAB ARTEnglishVibecheck explores methods for managing humor and tone within social platform interactions. This project demonstrates how automated systems can be applied to community moderation and content curation to maintain specific engagement standards.
JEVLAB ARTEnglishThis macOS utility reimagines window switching by integrating artificial intelligence to predict user intent. The project demonstrates a novel approach to desktop navigation, though specific performance metrics regarding its predictive accuracy remain unverified by independent testing.
JEVLAB ARTEnglishTool for systematic reviewers to extract information from research articles efficiently.
JEVLAB ARTEnglishOpenvons (open-Jev): 有限選択肢に確率で答える判断層 — テキスト / 画像 / 日本語音声コマンド.
JEVLAB ARTEnglishMulti-axis writing quality checker powered by TypeSafe AI's Jev model. Separate named checks, each with its own verdict and confidence.
JEVLAB ARTEnglishThis tool integrates Jev-powered ranking into zsh shell history. It demonstrates how to provide Fish-style autosuggestions by leveraging TypeSafe AI to prioritize command history based on user context and relevance.
JEVLAB ARTEnglishA curated list of awesome Jev / TypeSafe System One applications, libraries, and resources.
JEVLAB ARTEnglishThis project demonstrates a camera-only autonomous drone simulation within the MuJoCo environment. It features a TypeSafe Jev judgment model integrated into the control loop, which the creator reports operates at a frequency of 2.5Hz.
JEVLAB ARTEnglish
Kevin WangLurk is a monitoring tool designed to track Reddit threads for AI-cited data. It provides automated analysis of threads, delivering insights directly to users through email, Discord, or Slack integrations to help monitor information usage.
EnglishJob Risk Analyzer provides a CLI and REST API interface to evaluate professional roles. It uses Jev to score occupations based on their potential exposure to AI-driven workforce displacement and their relative resilience in an evolving labor market.
JEVLAB ARTEnglish
Aditya SinghThis project demonstrates a voice-controlled interface for Fusion 360. It uses Jev to distinguish between actionable commands and casual conversation, aiming to streamline CAD workflows through natural language input.
EnglishBenchmarks and a playground for TypeSafe's Jev (System One) model: chess, and who-is-the-player-talking-to for speech-to-text game NPCs.
JEVLAB ARTEnglishTypeSafe AIThis resource defines the JsonValue type alias for the TypeSafe AI SDK. It demonstrates a recursive TypeScript structure designed to represent any valid JSON-compatible data, including strings, numbers, booleans, nulls, arrays, and nested objects.
JEVLAB ARTEnglishA community post describing control of a simulated fly.
JEVLAB ARTEnglishA guide exploring Jev's architecture through 10,000 API calls.
JEVLAB ARTEnglishA tool for automated risk review of pull requests using Jev.
JEVLAB ARTEnglishTypeSafe AIThis TypeSafe AI guide explains the autoresearch feature discovery process. It demonstrates how automated systems can identify and categorize new features within a codebase to improve development efficiency and maintain project documentation.
JEVLAB ARTEnglishThis repository provides a skill set for developers to write and refine programs that interact with Jev, the System One model from TypeSafe. It serves as a practical guide for integrating model calls into software workflows.
JEVLAB ARTEnglishThis project demonstrates the application of Jev to predict human move patterns in a game of rock-paper-scissors. The creator claims the system identifies behavioral trends to anticipate player choices during competitive play.
JEVLAB ARTEnglishTypeSafe AIThis resource provides the foundational documentation for TypeSafe AI. It covers core primitives, implementation patterns, and practical cookbooks, alongside comprehensive technical references for the HTTP API and available software development kits.
JEVLAB ARTEnglishJev integrates with Netlify's AI Gateway for zero-configuration use.
JEVLAB ARTEnglishThis project demonstrates an automated agent designed to identify and filter low-quality content as users scroll through their feeds. It aims to improve digital consumption by providing real-time classification of incoming information streams.
JEVLAB ARTEnglishThis project introduces a System One-style model fine-tuned from Qwen3.5-2B. It demonstrates a method for performing single-pass typed decision-making tasks while incorporating calibrated probability outputs for improved reliability in automated reasoning workflows.
JEVLAB ARTEnglishSQL with natural-language predicates, powered by TypeSafe's Jev. Filter, rank, classify and score rows by meaning — batched, cached and cost-guarded.
JEVLAB ARTEnglishThis project provides a side-by-side comparison of TypeSafe and DeepSeek-flash. It evaluates performance metrics including speed, token usage, cost, and accuracy across tasks like invoice extraction, email classification, and reranking.
JEVLAB ARTEnglishThis experiment demonstrates a Jev-style typed-decision interface using a frozen Qwen3-4B model. It teaches how to extract option letter logits directly instead of relying on standard JSON generation for decision-making tasks.
JEVLAB ARTEnglishAn optimized inference engine to turn LLMs into Jev-like machines: optimized for quick, lightweight, and accurate decision-making, classification, and scoring.
JEVLAB ARTEnglish
Submit a find
한국어
日本語
Español
Français
Deutsch
简体中文
繁體中文
Português
Bahasa Indonesia
ไทย
Italiano
Tiếng Việt
Türkçe
Polski
Русский
4 HIGHLIGHTS
1
Agents & browsers
Tools & apps
Research & data
Start here
Decisions & routing
Games & real time
Devices & robotics
Creative & growth
Markets & experiments