The Jev launch post
Diogo Almeida introduces Jev, TypeSafe’s structured decision model. Watch the launch film and explore the original thread for the model’s approach, examples and author-reported performance.
Latest addition
Source check
Recent activityA FIELD GUIDE TO JEV / VOL. 01
The best of Jev. Real projects, practical guides and ideas from across the internet.
Diogo Almeida introduces Jev, TypeSafe’s structured decision model. Watch the launch film and explore the original thread for the model’s approach, examples and author-reported performance.
TypeSafe’s official launch announcement introduces the lab and points builders to Jev. It includes Diogo Almeida’s launch film and original introduction.
TypeSafe announces public access to Jev without a waiting list. The official console is the starting point for trying typed decisions in your own application.
TypeSafe highlights Jev’s typed-output approach: applications provide state and questions, then act on probabilities, choices or scores. The linked announcement also covers the Venice API integration.
Start with the Launch Post. Explore four official signals.
94 resources
AI-assisted summaries and translations. Check original sources for context and performance claims.
Demo showing Jev selecting the most suitable AI models for video and image generation tasks on Higgsfield.
Source-linked curation
Classifier that assesses prompt difficulty in real-time, enabling fast mode for simple prompts to enhance user interaction.
Source-linked curation
A resource detailing a test of Jev on 12 real-world scenarios, including speed, cost, and practical applications.

Source-linked curation
Jev enables flexible, type-safe classification as a fundamental programming primitive.
Source-linked curation
Analyzes 100,000 X posts in 20.4 seconds using 14 yes/no questions per post for viral potential assessment.
Source-linked curation
A tool that scores video ad shots in 1.5 seconds using Maxfusion and Jev, processing over 450 ads in under three minutes.
Source-linked curation
Runnable Jev examples through OpenRouter for immediate implementation.
JEVLAB ARTSource-linked curation
DuckDB extension that classifies rows in CSV, Parquet, or DuckDB tables with Jev, about 10 seconds per 1,000 rows.
Source-linked curation
A guide explaining Jev's approach to making typed decisions rather than processing text.
JEVLAB ARTSource-linked curation
A project classifying 1,018 AI research papers using Jev, achieving low cost and fast processing for organizing academic work.
Source-linked curation
A task using Jev to score 3,000 kids' snacks with multiple criteria in 28 seconds.
Source-linked curation
A real-time computer assistant using Jev and local Whisper to process voice input, classify actions, and interact with the screen through a Swift app for accessibility.
Source-linked curation
A setup where Jev selects between different AI models for specific tasks, streamlining model usage in a single terminal.

Source-linked curation
A Jev-powered system that analyzes PDFs page by page to determine which require OCR processing.
Source-linked curation
A project using Jev for skill routing in an agent system, based on community development.
Source-linked curation
An image classifier using OCR and Jev processes 900 images in 40 seconds.
Source-linked curation
An open-source project that helps select relevant Claude Code skills by prioritizing frequently used ones through Jev integration.
Source-linked curation
Jev is a System One model on OpenRouter that provides decisions with probabilities.
Source-linked curation
A GitHub project routing tasks by selecting model tier, tools, skill, and effort.
JEVLAB ARTSource-linked curation
Jev classifies 500 emails quickly and inexpensively, demonstrating its efficiency in processing tasks.
Source-linked curation
Classifies tax documents using an LLM pipeline for efficient processing.
Source-linked curation
Uses Jev to decide which AI model to use before generating output, based on prompt and quality requirements.
Source-linked curation
A GitHub Action using Jev for automated PR triage and classification.
JEVLAB ARTSource-linked curation
Jev evaluates the pacing of a frontier-AI essay written by @DarioAmodei.
Source-linked curation
Jev enables Grok Bot to control real Chrome for faster, automated tasks instead of slow look-and-click methods.
Source-linked curation
Turn any open model into a classifier/jev endpoint.
Source-linked curation
1,800-point thread debating whether typed decisions replace LLM calls for classification, routing, and scoring.
Source-linked curation
GitHub Action for issue triage that abstains: label, spam, needs-info and duplicate in one call, each applied only above a threshold you set, and nothing at all below it.
Source-linked curation
Pre-alpha PostgreSQL extension for TypeSafe AI (Jev) categorical classification.
JEVLAB ARTSource-linked curation
Real-time Discord moderation bot: Jev evaluates messages and metadata in parallel to catch phishing, spam, and social engineering with a progressive escalation ladder.
JEVLAB ARTSource-linked curation
Jev is demonstrated as a classification model with practical examples.

Source-linked curation
Open-source Jev-style System One decision model. Gemma 3 270M with a scoring head — fast, calibrated decisions in a single forward pass. No text generation. Inspired by TypeSafe.ai's Jev.
Source-linked curation
TypeSafe (Jev) vs DeepSeek-flash: side-by-side speed/token/cost/accuracy comparison across invoice extraction, email classification, and reranking.
Source-linked curation
A developer project for lead scoring in Clay using Jev.
Source-linked curation
jev-router (skill). A developer project using Jev to select appropriate models for prompts in coding tasks.
Source-linked curation
A GitHub project for a Pareto-optimal OpenRouter router for Pi, decided by Jev.
Source-linked curation
A prototype that classifies symptoms and updates diagnoses using medical ontologies during live clinical consultations.
Source-linked curation
Comment-moderation playground: paste a comment, Jev decides what to do with it.
Source-linked curation
A GitHub project that uses Jev to triage and route posts from the live Bluesky firehose with human oversight.
JEVLAB ARTSource-linked curation
A 26-sheet construction plan set was classified in 2.9 seconds using Jev with an LLM pipeline, costing $0.0052 and matching GPT-4.1 and GPT-6 Astra performance.
Source-linked curation
A community post about using Jev for incident triage in Box, returning boolean results with probabilities.
Source-linked curation
A community post presents a codebase complexity classifier using Jev, aiming to address overengineered code.
Source-linked curation
A guide on using Jev to route between models and block risky tool calls.
JEVLAB ARTSource-linked curation
A community example demonstrating how Jev can classify company invoices into accounting categories rapidly.
Source-linked curation
A demonstration of Jev's capability to quickly categorize a large volume of Hacker News posts.
Source-linked curation
Screen a folder of CVs with the TypeSafe Jev decision model: typed judgments, an editable policy, free re-scoring.
Source-linked curation
A demo showing how to make open source models behave like Jev using inference engineering and scoring endpoints for decision-making.
Source-linked curation
SuperX integrated with Jev enables rapid post scoring, analyzing 61 questions per post in under a second with high viral potential prediction accuracy.
Source-linked curation
Classify Git commit diffs and messages with Jev. Bug fixes, security fixes/CWEs, and change types.
Source-linked curation
Calibration and confidence-based routing measured on Banking77: 80.2% accuracy at $0.103 per 500 decisions.
Source-linked curation
A developer project exploring decisions using a classifier as a next-token predictor.
Source-linked curation
Guardrails for LLM apps in one API call. Prompt injection, jailbreaks, leaks, unsafe content. Built on TypeSafe Jev. MIT.
Source-linked curation
Eight minimal working examples of TypeSafe's Jev (a System One model) applied to mechanical and electrical engineering: CAD/CAE/CAM routing, FEM result triage, DFM screening, BOM alignment, hallucination-proof extraction. Zero dependencies.
Source-linked curation
Neon Function proxy for the Neon AI Gateway with TypeSafe Jev routing.
Source-linked curation
Jev demonstrated over 5x faster performance than other models in testing.
Source-linked curation
A community post detailing a dashboard that classifies X posts across eight dimensions using Jev for content research.
Source-linked curation
Self-hosted AI email classifier for Gmail powered by Jev. Create custom labels, organize your inbox, and filter spam with confidence and cost controls.
Source-linked curation
Helper n8n community node for Jev by TypeSafe. Classify, route, and score text with questions you define, and get a probability for every answer so unsure items can go to review.
Source-linked curation
A system tested on 1,500 emails for classification effectiveness. Demonstrates potential for automated email management.
Source-linked curation
An ad blocker extension using AI for real-time classification and removal of ads.
Source-linked curation
Marketing analytics platform whose feature flag routes brand-visibility classifiers off an LLM and onto Jev boolean decisions.
JEVLAB ARTSource-linked curation
Short practical intro with a Python ticket-triage example.
Source-linked curation
Experimental semantic line search with TypeSafe Jev via OpenRouter. Python CLI with no runtime dependencies.
Source-linked curation
Autonomous System-One Triage Engine & Benchmark powered by TypeSafe AI (Jev). 75ms inference, $0 output tokens, and RLCD epistemic safety gates.
Source-linked curation
A developer project for model routing with caching and failover.
Source-linked curation
Jev is utilized for decision-making processes on Cloudflare, including support routing and risk management.
JEVLAB ARTSource-linked curation
Open-source Jev log triage for OpenTelemetry. Score the signal before expensive LLM analysis.
Source-linked curation
Jev enables natural language search for Zillow listings, classifying properties by non-standard filters.
Source-linked curation
GitHub project for content moderation with category-specific probability thresholds. Enables automated content filtering.
JEVLAB ARTSource-linked curation
LaneBreak — support ticket priority+routing via TypeSafe Jev.
Source-linked curation
A LangChain package with TypeSafeClassifier for integrating Jev into applications.
Source-linked curation
JEV Document Classification enables the rapid and cost-effective classification of text-based documents using AI, leveraging TypeSafe's "System One" model.
Source-linked curation