Projects built with Jev

No projects found. Clear filters

Design decisions without text generationAs suspected. ~$0.0005 a design. It flies. What if the model designing your UI can't write a single word? Well, it seems like you don't need words like a lot of harness practices we got accustomed to this year. Using Jev, a tiny decision model, picks the whole design in ~1Paul ⏳

Jevscan flags suspicious Etherscan activityBuilt jevscan - a Chrome extension to detect malicious addresses and transactions on @etherscan in real-time using Jev from @typesafeai.Raz0r

An npm package gate for coding agentsi used Jev to decide whether a coding agent should ingest an npm package. paste a name. fetch the real README and agent files. jev returns allow or refuse. code does the rest.Hatif

Phone automation with step verificationFinally got access to Jev, so I gave it a phone. Plain English in: sign in, add a ₹900 Travel expense, turn on dark mode. LLM made 10 steps once. Jev handled every step twice: what to tap + did it work. 10/10. p=1.00. ~400ms each. Entire run: $0.0006.Rudra

A live market-news research deskThe whole build cost under a dollar in model calls. A live market news desk: yfinance headlines in, ranked trade ideas out. 215 typed judgments per pass, 2.7 seconds, a quarter of a cent. Claude Code + TypeSafe's agent skill. #jev @typesafeaiNairb

Fifty Wordles at oncejev played 50 wordles at the same time. 49 wins. 3.78 guesses average. 15.8 seconds. 4 cents. the one it lost was TATTY. it burned 5 guesses on tangy, tacky, taffy, tally, tabby. it got trapped exactly like a human does. peak AGIAhmed Gagan

Worth Replying finds relevant X discussionsWe built "Worth Replying" with Jev + AIsa. Drop in a company’s domain and it finds people on X already talking about the problems your product solves. We tested it on : > 150 tweets found > 750 Jev decisions in 18.8s > 79 worth replying to > 71 archivedAlsa

Tenet review gates for agent-written codeSpent a day building tenet, a review gate for code that agents write. It judges each commit, commit message and PR against rules written in plain language, using Jev by @typesafeai. The rules come from @poteto's unslop skill, @hvpandya's stop-slop, @dillon_mulroy's anti-slop andTim Cheung

Native Rust network-traffic monitoringWhat if an AI watched your computer’s traffic and told you when something looked unusual? Couldn’t do it with an LLM. Too slow. @typesafeai’s Jev is fast enough. I wired it into a native Rust app. Built a Rust SDK while I was at it. 🦀Phillip

Ranking relevant Reddit discussionsJev can’t write a single sentence but sorted Reddit for me. 12 hours with Jev: 180k+ threads collected 12k in target subs 205 worth replying to, ranked by relevancy Self-improving growth loop. Cost: free.Johan Cutych

Filtering unwanted X repliesPro tip: You can use Jev to remove annoying reply guy comments that X seems to always miss. Takes 5 minutes in Astra with the docs and an API key.Ian Nuttall

A document jargon detectorWanna check your documents for corporate, legal or technical jargon? I played around with this and made a jargon detector. Paste in any document and it instantly shows you where there is jargon. Super fast and more or less for free. It's just slicing the document in sentencesValentin Sattinger

Jev plays Clash RoyaleMade Jev play Clash Royale and actually won a match.Amrit Nigam

Parallel adversarial browser testingthanks to @typesafeai's Jev we now have massively parallel browser-based adversarial testing suite that tries to break each release. and it costs pennies.Rafal Wilinski

Paper Breakdown research recommendationsOrganically integrated JEV in a real project The Paper Breakdown Recommendation Engine We already had a working system that combines content-based and collaborative filtering. Pro users may now get additional curation with Jev. 0.0019$/user for ~70 recs 🤯 ~50 users in 5 secsAVB

Judgment cache vs semantic cachegot access to @typesafeai Jev, and my first use case was building a judgement cache. I compared it with a semantic cache. One uses cosine similarity with a >0.90 threshold. The other uses Jev.ᴮᴷ

Jevton: a town of 120 peopleWTF !!! I pushed @typesafeai to a break neck decision making frenzy and it held up its own. Insane I built a town where Jev makes EVERY decision and never writes a single word 🤯 Jevton 🏘️ 120 people, 7 roads, 20 businesses, a council, a newspaper Every hour Jev decides forChizi

A smart calculator notebooki made a smart calculator notebook using jev it can calculate ANYTHING!! 😅kitze 🛠️ tinkerer.club

Slay the Spire 2 in real timeI couldn't think of something original so here's Jev playing Slay the Spire 2 in realtime (1X speed). It gets through Act 1 in 9 minutes and cost just $0.015. The speed here is just incredible. It's like watching a speed runner. It didn't always make the best decisions andPranay Prakash

Scoring every shot in a video adJev is insane for ad research. I built a tool that scores every shot of any video ad on meta in 1.5 seconds. It runs on Maxfusion + @typesafeai I pushed 450+ ads through it in less than 3 minutes. It pulls any advertisers live video ads from Research Lab on Maxfusion.Vlad Dubchak

Instant decisions inside a SaaS agentAs an experiment, we plugged Jev by @typesafeai into the AI Agent of our SaaS. The speed is absolutely incredible. It answers faster than I can type or click and costs ~100x less than a classic LLM. 🤯 And when a request still needs a full LLM, it tells us in a third of aLennart

A function that evaluates the vibe@typesafeai Jev is really cool. I recorded a quick demo of a vibe function that evaluates the vibe of an input. I can already see a ton of uses for Jev in complex systemsMonocursive

Jev Trades market simulationI gave Jev $100,000 to trade on ethereum:native and to my surprise it made a profit of 0.28% in 15 minutes. Check it out : GitHub repo : Let’s #connect if you’re in fintechZade

A Jev-controlled cityJev + GPT-6 Astra = INSANE GPT-6 Astra built this entire city from ONE prompt. Then I gave control of it to Jev. I wanted to test what happens when you combine a powerful coding model with a model designed to make tiny decisions extremely fast and cheaply. So I asked Astra toMaxim Oulitzky

Naruto street fighting against Jevmade a Naruto-themed street fighter game and played against @typesafeai’s Jev it made ~130 decisions in a 50s fight, barely registered any usage, and came very close to beating me. very impressed by its reaction time and use of special move (substitution jutsu).Kausthub Jadhav

Orus trade-strategy reviewAs promised, Jev is now integrated into Orus agents. What is Jev, and how can it help our agents? Jev is a model from TypeSafe AI, built to answer precise questions with structured answers. In Orus, it reviews the trades your strategy proposes, before they happen. It approvesOrus Agent

Predicting the next shell commandJev picking the next command from shell historyThorsten Ball

Instant calorie trackingi made a calorie tracking app with jev. it's INSTANT 🤯kitze 🛠️ tinkerer.club

A semantic circuit breaker for agentsJev unlocked a new way to stop AI agents from spinning. Agents do not always repeat the same action. Sometimes they try different things based on the same wrong assumption. I built a semantic circuit breaker around that. In one stuck run: 12 model calls -> 4 19.9K tokens ->Ashutosh Mathore

Synkora resume-matching experimentI’ve been testing @typesafeai JEV (TypeSafe) with Synkora. I gave Claude Sonnet access to JEV as a tool and tested it against 5 resumes for a single job profile. The result: 5 resumes evaluated Structured evaluation through JEV Each evaluation completed in milliseconds TheRaju Mazumder

RADAR ranks relevant developer resourcesdevelopers keep asking “has anyone solved this?” so I built a test for @typesafeai Jev: RADAR finds the repos, packages, issues + papers jev ranks what’s relevant / actionable / worth opening first Codex one-shotted the whole test and came up with the idea first case: passkeysTedOS

A robot arm picks up a ballI gave Jev a robot arm. Two USB cameras, an SO-101, and @typesafeai's Jev choosing the next move at ~2 decisions a second. After a slip and some debugging, it went back for the ball and got it into the bowl.Daniiar Abdiev

Jevcal confidence-threshold calibrationEveryone picks Jev confidence thresholds by vibes. 0.95? 0.5? I built jevcal. Give it your data and say "I need 99% accuracy". You get the exact threshold, how much Jev can handle, and how much still needs an LLM. Open source, one command. @typesafeaiAbhishek kothari

AIBackends Jev API integrationAIBackends web API now supports Jev by @typesafeai If you need some abstraction from your main app where you can customise the integration, you can use this API layerMelvin Vivas

A job-description roasting machineI built a job-roasting machine on Jev - paste a JD, it grills it line by line. It's insane how cheap and fast it is at scoring! @typesafeai "We're like a family here" → Boundaries are not a thing here. "Wear many hats" → You are 3 employees. "Comfortable with ambiguity" →Thilak Bhat

Jev attempts Resident EvilSo @typesafeai Jev knows how to shoot and dodge, but can Jev survives against zombies in 3D space??? I had Jev played the original Resident Evil to see how far Jev could go. This was an extreme case. 3D space, understanding objectives, fight off enemies, doing puzzles, etc.Jack Vinijtrongjit