X posts

SEO AIO GEO Dashboard built with Jev. https://t.co/q6pcGY9ouh

There's a new AI model that refuses to talk. While everyone debates what a "decision model" is even for, builders already shipped the answer. 7 things people built with Jev in 9 days. Number 5 broke my brain 👇

made a memory system using jev; results are very promising so far, stay tuned for the repo! #Jev #claude #gpt #git #ai

Had fun figuring our Jev as deterministic tool to aid LLM as a judge (gpt-6). Synthetic simple data use case using jev. It’s so cheap running it. https://t.co/tQWE2HD4A6

I loved this idea so much that it deserves a share ❤️ I recently asked people to share what they’ve built with JEV, and then this guy 👤 shows up with an entire website dedicated to different JEV use cases! That’s honestly pretty cool. If you’re curious about JEV and want to see how it can actually be applied, definitely check this out. Huge shoutout to Animesh Mishra (@thatcasualvc) for the wonderful share 🙌 This is exactly the kind of stuff I love seeing from builders. 🚀

npm install is the scariest command in your terminal. One typo → lookalike package → malicious postinstall → secrets gone. jev-sec-audit flags it in milliseconds using Jev, a System 1 model built for fast decisions, not chat. One step in GitHub Actions. Open source. https://t.co/PhC7TLRwCt

Jev 是怎么打 DOOM 的？ 它读游戏导出的结构化状态，选择目标和行为，再由控制器执行具体操作。把“司令官”和“手脚”分开，才看得清这个演示是怎么搭起来的。 做了一期 5 分钟拆解，讲清模型、动作抽象和控制器各自负责什么。#Jev #GameAI

Jevでファイル名や変数名じゃなく「何をしてる場所か」で探せるVS Code拡張を作った。 「FVのフェードインアニメーション」「CMSにアクセスしている箇所」とかで検索可能。 claude code でもちゃんと最速モデルを選べば5~10sくらいで出るが、この拡張機能（Jev）なら0.5sでほぼリアルタイム。

The most talked-about AI model this week is Jev, released by TypeSafe AI. Jev is designed to make fast, structured decisions. I thought career comparison would be a really interesting way to test it. So I used Jev to build this website, It helps you compare your current job with the role you want, based on what actually matters to you, not just the job title or salary. It creates a career card with your name, both scores, and your Ikigai Gap. You can download it, or share it with someone who's also thinking about a career change. If you're thinking about your next career move, or you just want

LLM をゲームで使ってて一番困ってたのが、考える時間の長さ。Agent 同士でリアルタイムに戦わせるのが、なかなか難しかったんですよね。 今回、@typesafeai の #Jev にアクセスできるようになったので、ゲームに組み込んで Sonnet Fast と動かしてみました。どんな感じかは下の動画を見てください！ 次は 5 秒のターン制限を外して、Laya と Jev をリアルタイムで戦わせてみようかな。

Built a tool that searches Google Icons using JEV (Laya-MLX) Should I publish it? 👀 https://t.co/97qw7oIkOt

best way to close my day with a webinar on @typesafeai Jev in JarvisCore! i covered the 4 places we are using Jev in our agent runtime: RAG classification and ranking, model routing, subagent routing, and directly in agent decisions this is not the end, i believe there is much more to come in our memory assembly, tool recovery, peer delegation, and human-in-the-loop. Every agent harness, runtime, loop needs some good Jevglue! also check out this https://t.co/nrzsIzpxM8 by @CompleteSkeptic

we're sponsoring https://t.co/QMtLnAp5i5, the gallery of everything built with jev by @kraayenjon. #2 on the board, right next to @openjevai . if you built something with jev, it belongs there. and on https://t.co/ac59bsQKFk. https://t.co/Bg1mMNCjnF

Built an AI slop finder for @X using Jev @typesafeai . • Tweet → Content script extracts text • Text → Backend API → Jev • Jev → AI-slop score • Score → Badge displayed on the tweet #jev #aislop #newmodels https://t.co/VMzJA1Tyde

Have been wanting to try Jev from typesafe AI all week. hopefully we will get some time to do so this weakend. What have you built with Jev? Info dump us please.

🤔 Where do #Jev and other "Decision Models" fit into your current agentic or non-agentic pipelines? The short answer: anywhere you might have LLM calls with a #pydantic model or json schema to type-cast your raw data into structured categories, Likert-scale numerical scores, or true/false decisions. I ran three benchmarks with @flyteorg on https://t.co/WczbNQoVLn in the context of three use cases: - Customer support - Code review - Legal contract review In all three cases, using Jev was faster (up to x2.6) and cheaper (at least x10) with no loss in quality compared to using a "System 2"

How we're using Jev frrom @typesafeai at @knowyourcompani - Rendering exact citation on a document page -> upgrade from chunk box mapping to more precise citations https://t.co/5pUBxNcLS7

This might be the most useful thing I’ve built with Jev so far. A Chrome extension that quietly analyzes what I’m seeing and classify them as Opinion, Engagement bait, etc... So clean you may think it is a new feature on X :) https://t.co/POUtFhxGkD

Jev all the things?? 🔥 built a CLI over the weekend that classifies an artist's entire discography by theme, mood, and lyrical complexity, using Jev (thanks @typesafeai) it then renders it as a terminal dashboard ran it on Nirvana. 52 songs, 1989–1993 pipeline is simple on purpose: → MusicBrainz resolves the artist + pulls the discography (free, no key) → https://t.co/w4qk26vniP fetches lyrics per track (also free, also no key) → Jev classifies each song against 5 questions in ONE batched call per track instead of 5 separate ones the "boring" part took longer than the Jev part tbh, get

What people actually built with Jev — a daily feed https://t.co/YBUNUrmAnU https://t.co/TT0zyPSQkx

I made an Algolia alternative using Jev https://t.co/USs2C6kKAS

BREAKING: This model is 400x cheaper and 200x faster than ChatGPT. It shipped 6 days ago. I scored 712 LinkedIn messages in 8.3 seconds. For 4 cents. Not with ChatGPT. With JEV, a model that writes nothing. It does one thing: decide. Yes or no, a score, a choice. With a calibrated probability. On 712 hot leads and 712 personalized DMs: → 7,120 typed scores (hook, offer, CTA, buying signal, ICP fit, intent…) → the best message picked for each lead → every lead/message mismatch caught before send → $0.042 per million tokens. $0.045 for the whole batch. The truth? Most of the time spent on outbou

Built a level design playtesting demo with Jev. 100 ghost players. 33 predicted to finish. 32 made it. The bottleneck? A laser window. Widened it, and 36 made it through. A simulated look at where players get stuck. Would you test your first level this way?

Customer Support is a pretty good usecase for @typesafeai's Jev Built a chat router, outputting verified pre-defined responses. the whole chat below cost $0.000083 total. 🤯 (1/2)

Opus 5.5 vs Astra GPT-6 creating a game called JEV pongs JEV This video is NOT AI generated! However, what you see here is a game where the new @typesafeai model JEV fights itself in a Pong clone game. See for yourself how Opus 5.5 blows Codex Astra GPT-6 out of the water. (Sound ON!) Want a deep dive video from first prompt to finish, let me know. Happy to show the details!

Everyone is talking about Jev, but there’s a new kid in town: Decision-Machine-1. So far in my tests, it’s faster, cheaper on estimated API cost, and beating Jev at tennis. 🎾 I wanted to understand these decision-making models, so I made it fun. I adapted a browser tennis game and let them play against each other. Now I’m watching API calls turn into rallies and wondering why one player keeps rushing the net. The players are Jev from @typesafeai and Decision-Machine-1 from @millisecondsai Each model gets a snapshot of the game state and picks a tactic. It can choose the shot direction and dep

I built JevSearch, search the web & validate your results with Jev. Give a query and selection criteria, use @browserbase search to get the t25 results, then Jev scores and returns the t5 results. Jev often chooses urls outside of the initial top 5 as more relevant.

been experimenting with a jev2mcp small demo using @typesafeai jev to decide which MCPs / tools / plugins should be injected into an agent prompt before it gets sent. it can handle messy prompts too. instead of making the agent figure out what tooling it needs, jev handles all the routing. pretty neat!

Opus5.5 x Cloudflare x jev メールの仕分け作業のサイトを作ってみました。 メール自体はダミーですが、仕分けはリアルにjevで動作させています。 テンション上がりまくりです。

been testing a few Jev models by making them play Pokémon & snake lol which is the best way I've found to see how a model handles a complex situation tested: Laya, @jaredpalmer's Kev, @flock_io This/That 1.1, @googlegemma djev

This is how you Jevmaxx! @typesafeai I built a chrome extension that scans your x feed for larp and hides it for you! The results are INSANE! Larp be gone! Your feed will be free of larp at under 250ms per post and for under $0.01 for your daily doomscrolling needs! Here's how I built this: 1/ Scan the entire available x post library 2/ Find patterns for classic larp indicators 3/ Train jev on the larp indicators 4/ Slop out a chrome extension This is how you fight larp with larp and make the world larp free Stay tuned for laya edition

I made a tool with Jev that reads my whole codebase to answer one question: what should I be looking at? Every chunk of code in your project is scored for relevance.

Claude Opus is insane! Sonic just won’t stop. Pair it with Jev, and I honestly can’t look away. 🤯 I built this on YouWare with Claude Opus 5.5 using one prompt: “Create a beautiful Sonic-inspired endless runner using Three.js and WebGL, with smooth controls, polished visuals and refined interactions. Deliver it as HTML.” The gameplay is so smooth. Play it yourself, or let Jev decide when to jump, roll and dodge. 10+ minutes in. Still running. Still watching. Try it Now with YouWare 👇🏻

Your elevator sucks. I gave Jev the keys. Jev-elator is a live elevator simulation where Jev decides the next floor—not your traditional dispatch algorithm. Jev is now responsible for balancing its obligation as an elevator operator with: - Prioritizing someone that needs to pee badly or is carrying luggage - An opportunity to generate revenue - Giving riders with bad karma a worse experience Did Jev perform better than the baseline? Try it: Disclaimer: ts slop 🥀

【衝撃】 同じ384件のニュースを 2つのAIに同時に読ませたら、 コストに390倍の差がつきました🔥 全部を高性能モデルに任せると、 出費も待ち時間も膨らむ実例です。 ・比べたのは判定AI「Jev」とClaude Opus 5 ・384件の見出しを15の企業ごとに振り分ける検証 ・Jevは24.9秒で384件を処理、約0.19ドル ・同じ時間でOpus 5は4件、約0.77ドル ・見出し1件あたりで比べると約390倍の差 投稿者のElvis氏は、この検証を PRエージェント向けOSSツール 「newsjack」のデモとして公開しています。 「大量の一次判定は専用の軽量モデルに、深い分析だけ汎用モデルに回す」 という設計が、コストと速度の 両方に効いている実例です。 Codexでエージェントを設計するときの、 タスクの粒度でモデルを使い分ける 発想の参考になります。 Codexも、1日に何回頼むかで判断すれば、 8割の人はPlusで足ります。 その選び方は 記事にまとめています👇 ============ 営業してもないのに大手企業からお願いされるくらいの、Codexのノウハウや最新情報を発信しています。 フォローして一緒にCodexをマスターしましょう！

天下武功，唯快不破 JEV很好，但是我选择laya Laya是一套开源的结构化判断模型 最大优势其实就是可以本地部署 可以魔改训练 比如BSC meme，我给他接了价格、曲线进度、流动性、买卖流、Top10 持仓、开发者持仓、聪明钱 再套到一个meme早盘流量轮动的策略里边 跑了一天，两个模型对比 jEV跑了200多笔，输17U Laya跑了312笔，胜7U 建议持续深耕高速决策模型 还有很深的应用场景可以挖掘

I don’t know if people still remember Jev. Things move so fast here. I’ve spent the last few days testing it, and found a few things I think builders should see. Prompt injection is one of the most interesting things to test in AI, so I made a little demo to show how it could change Jev’s answer. Jev is a great model. I just want people to know what to watch out for when they use it.

BlockRun × Jev ⚡ Introducing SignalDesk. We gave Jev 100 real X posts. It made 300 intent + product-fit judgments in 3.9 seconds. Find people asking for alternatives, see if your product fits, and get a reply draft to review. Open source. Demo below.

Jevを使ったアプリを作ってみた。 コピーしたテキストをペースト先の文脈に沿った形でペーストされていくシンプルなクリップボードツール。もちろん既存のクリップボードの機能もある。 普通に便利だったので公開する準備してる。

Tested Jev (@typesafe_ai) paired with Claude for the first time, and the results for programmatic evaluation are impressive. 🧵👇 Takeaway: Don't force general LLMs to handle deterministic scoring. Pairing Claude’s/Codex reasoning What use cases would you test this on?