Skip to content

Case wall

Jev AI examples, page 4

615 real Jev AI builds and public demos, updated daily. Original text, metrics, and a link back to each source.

Last updated

Public Jev AI examples and demos, not official TypeSafe docs. Clients and SDKs are in the Jev GitHub projects. Paths for jev-latest, Vercel, and OpenRouter are in the September 2026 channel map.

Agent-skill repositories are listed on Jev skills on GitHub.

OpenRouter model ids recorded for Jev AI are on Jev AI on OpenRouter. Rubric and eval posts are on Jev as a judge. How the company and the model fit together is on TypeSafe AI. The pricing calculator uses the published input rate.

Filter

24

Celeris-1.1 Decision recorded the lowest Time per Task at 56 milliseconds, followed by Elephant 9B (70 milliseconds) and Jev 1.14 (92 milliseconds). In comparis…

Benchmarks

Artificial Analysis @ArtificialAnlys

EN

@ArtificialAnlys reports Decision Index time per task: Celeris-1.1 Decision 56 ms, Elephant 9B 70 ms, Jev 1.14 92 ms, and Claude Haiku 5.5 (Max) 32.65 seconds. Mammoth 27B scores within 4 points of GPT-6 Luna (Max) and is 118× faster.

Celeris-1.1 Decision recorded the lowest Time per Task at 56 milliseconds, followed by Elephant 9B (70 milliseconds) and Jev 1.14 (92 milliseconds). In comparis…

Poster from Strawberry (@strawberry)

Apps

Strawberry @strawberry

EN

@strawberry introduces Jev-powered app triggers: monitor thousands of app events for cents so companions keep working. Examples include watching Gmail for support mail and waiting for a Slack reply before filing an NDA in Drive. Jev classifies incoming events so the companion continues until the task is done. Free trial at strawberrybrowser.com.

Introducing Jev @typesafeai powered app triggers No more babysitting your agents, monitor 1000s of app events for cents Try it for free at http://strawberrybrow…

Poster from Monid (@MonidHQ)

Apps

Monid @MonidHQ

EN

@MonidHQ launched Jev-powered viral UGC discovery that scores TikTok/Instagram/YouTube creators from their last 50 posts for about $0.002 each, aimed at agent workflows.

Introducing @typesafeai Jev-powered Viral UGC Discovery Find high-performing creators for just $0.002 each. No more $199/mo influencer tools. Your AI agents can…

Poster from Antoine 🤖 (@antoinemarcel)

Apps

Antoine 🤖 @antoinemarcel

EN

@antoinemarcel demos fully autonomous clothing pick-and-putaway: SLAM + Jev locates garments with 97% success on unseen items.

Fully autonomous demo: Pick up and put away any piece of clothing. 97% success rate on unseen garments. Stack: SLAM + JEV to locate the clothes. Astra + PI0.5 +…

Found a really cool use-case for jev from @typesafeai. I have a custom slack connector for our agent Gilles, and any time a thread is responded to, we hand it o…

Agents

marston connell @MarstonConnell

EN

@MarstonConnell has a custom Slack connector for their agent Gilles. Whenever a thread is responded to, they hand the message to Jev from @typesafeai to check whether it was meant for the agent, so they do not spam the agent in large threads.

Found a really cool use-case for jev from @typesafeai. I have a custom slack connector for our agent Gilles, and any time a thread is responded to, we hand it o…

What a fast, cheap and reliable judgment model has in common with humans and where it differs. Jev from @TypeSafeAI can support useful cognitive tests at a scal…

Guides

Alexander Pico @xanderpico

EN

@xanderpico writes that a fast, cheap and reliable judgment model shares some traits with humans and differs in others, and that Jev from @TypeSafeAI can support cognitive tests at a scale that used to be impractical.

What a fast, cheap and reliable judgment model has in common with humans and where it differs. Jev from @TypeSafeAI can support useful cognitive tests at a scal…

Poster from Patrik Nordin (@PatrikNordinHR)

Agents

Patrik Nordin @PatrikNordinHR

EN

@PatrikNordinHR says Grok Bot from @cursor_ai coordinates the work, while @claudeai and Codex from @OpenAI work on the tasks together. Jev from @typesafeai helps decide which agent should take the next step. Reason Network from @reasonmachines passes the task and its context, so agents can be switched without briefing the next one from scratch.

Grok Bot from @cursor_ai coordinates the work in my AI setup, while @claudeai and Codex from @OpenAI work together on the tasks. I use Jev from @typesafeai to h…

Jev is now a native connector on Voximplant. Your code can ask it whether the caller finished the thought or actually agreed to the booking, and get a probabili…

Infrastructure

Voximplant @voximplant

EN

@voximplant says Jev is now a native connector, so code can ask whether a caller finished a thought or actually agreed to a booking and get a probability back in a few hundred milliseconds.

Jev is now a native connector on Voximplant. Your code can ask it whether the caller finished the thought or actually agreed to the booking, and get a probabili…

our eng team got #1 on HN today! they tested if fast + cheap decision models like Jev @typesafeai could grade complex data-matching tasks as well as frontier mo…

Benchmarks

Tejas Manohar @tejasmanohar

EN

@tejasmanohar says Hightouch’s eng post hit #1 on HN: Jev+Opus nearly matched Opus alone on hard data-matching grades at much lower cost.

our eng team got #1 on HN today! they tested if fast + cheap decision models like Jev @typesafeai could grade complex data-matching tasks as well as frontier mo…

An AI model KNOCKED ME OUT! Tried building a Boxing game using Jev to see if it can be a good adversary, and you can do it too. I walk you through the project i…

Apps

Arindam @arindamkeswani

EN

@arindamkeswani says an AI model knocked him out. He tried building a boxing game that uses Jev as the adversary, and says you can do it too. He walks through the project in a YouTube video.

An AI model KNOCKED ME OUT! Tried building a Boxing game using Jev to see if it can be a good adversary, and you can do it too. I walk you through the project i…

Prompt injection in the wild: a Youmake user tried to access credentials through a malicious prompt. Our guardrails, built with JEV from @typesafeai, successful…

Apps

BABA Khalid | Youmake.dev @BabaKhaliid

EN

@BabaKhaliid says a Youmake user tried to access credentials through a malicious prompt. Guardrails built with Jev from @typesafeai blocked the attempt.

Prompt injection in the wild: a Youmake user tried to access credentials through a malicious prompt. Our guardrails, built with JEV from @typesafeai, successful…

Poster from William Zhu (@WilliamZhu3541)

Apps

William Zhu @WilliamZhu3541

EN

@WilliamZhu3541 introduces SavorLM, where saying "French, near my hotel, open now, under $70" gets three restaurant picks from a raccoon named Savvy, and he lists @typesafeai Jev among what it was built with.

I have a confession to make. I can't resist a great French restaurant, especially the duck confit. But every time I land in a new city, I spend 40 minutes on Go…

Poster from Vitaly Kleban (@vkleban)

Apps

Vitaly Kleban @vkleban

EN

@vkleban built https://jevfight.com, a fighting game where both players are controlled by Jev models. He says to tell him your fighter name and he will attack you.

I've built https://jevfight.com - fighting game where both players are controlled by Jev models! 🥊 Tell me your fighter name and I will attack you!

Poster from Kaif (@kaif9998)

Agents

Kaif @kaif9998

EN

@kaif9998 used Treg to pull 20 niche YouTube videos and Jev to sort the top 10 into opinion/demo/tutorial for about $0.00112.

MY AI AGENT'S RESEARCH BUDGET: $0.00112 Treg pulled 20 YouTube videos in my niche. Jev sorted the top 10: opinion, demo, tutorial. Top opinion video: 319K views…

I accidentally made a Bloomberg for Pokémon cards… Bill Benter made about $500 million betting on horses and two years ago it inspired us to start building a st…

Apps

Smoke the Books @SmokeTheBooks

EN

@SmokeTheBooks built a “Bloomberg for Pokémon cards” after cleaning ~80 million sales records. After normal filtering, they use Jev from @typesafeai to classify millions of difficult eBay listing cases. The site has screeners, portfolio tools, and an MCP server; sign-in required for predictions.

I accidentally made a Bloomberg for Pokémon cards… Bill Benter made about $500 million betting on horses and two years ago it inspired us to start building a st…

The @convex Discord bot is a special cookie, Ian built a strong foundation for it two years ago. We updated it recently, added Jev @typesafeai, and bam, we got…

Agents

Sara @_sarant

EN

@_sarant says the Convex Discord mod bot now uses Jev for moderation decisions and costs less than 10 cents, updating a bot Ian started two years ago.

The @convex Discord bot is a special cookie, Ian built a strong foundation for it two years ago. We updated it recently, added Jev @typesafeai, and bam, we got…

Poster from Shrivu Shankar (@ShrivuShankar)

Apps

Shrivu Shankar @ShrivuShankar

EN

@ShrivuShankar says the latest Infinite Alchemy update uses @typesafeai's Jev so players can build cheap, just-in-time semantic sorts and groupings over thousands of discovered elements.

snuck in @typesafeai 's jev into the latest update of infinite alchemy as you play the game, it's hard to be organized with 1000s of 'discovered elements'. usin…

On a generic decisions benchmark, OpenAI, Jev, and Parallel all score about 90%. On questions that depend on the live web, OpenAI drops 2.2x, and Jev drops 2.6x…

Benchmarks

Parallel Web Systems @p0

EN

@p0 (Parallel Web Systems) says on a generic decisions benchmark OpenAI, Jev, and Parallel all score about 90%, but on live-web questions OpenAI drops 2.2× and Jev drops 2.6×. They built PWS-Decisions so generic benches cannot hide that gap, and link the benchmark blog.

On a generic decisions benchmark, OpenAI, Jev, and Parallel all score about 90%. On questions that depend on the live web, OpenAI drops 2.2x, and Jev drops 2.6x…

One thing that can be tricky to figure out is how to use Jev's primitives to handle more than one intent in a single user query. My favorite way to approach thi…

Guides

Allie the Icon @allietheicon

EN

@allietheicon describes a "shopping cart" pattern for more than one intent in a query: recursively pick the next thing until Jev says done. She used it for a Slack coffee bot (DoorDash CLI) and a smart-home handler.

One thing that can be tricky to figure out is how to use Jev's primitives to handle more than one intent in a single user query. My favorite way to approach thi…

I released Kkdev92.Jev v0.2.0-alpha! 🔀 My unofficial .NET SDK for TypeSafe Jev now supports PNG/JPEG images, experimental Box questions for locating objects, an…

Apps

kkdev92 @kkdev92_dev

EN

@kkdev92_dev released Kkdev92.Jev v0.2.0-alpha, an unofficial .NET SDK for TypeSafe Jev that now supports PNG and JPEG images, experimental Box questions for locating objects, and long context.

I released Kkdev92.Jev v0.2.0-alpha! 🔀 My unofficial .NET SDK for TypeSafe Jev now supports PNG/JPEG images, experimental Box questions for locating objects, an…

Poster from Jakub Chrzanowski 🥑 (@hszanowski)

Apps

Jakub Chrzanowski 🥑 @hszanowski

EN

@hszanowski shows an early prototype that explores natural-language “what’s this called?” lookup in Search Everywhere using Jev and the Decisions API. Asks what queries people struggle to find in their IDE.

“Surely my IDE can do this. What’s it called, though?” We’re exploring that with Jev and the Decisions API in Search Everywhere. Here’s an early prototype. What…

Two Jev passes helped our Graph RAG pipeline outperform the LLM rerankers we tested, with higher recall and less time spent on model calls. Earlier, we released…

Benchmarks

Milvus @milvusio

EN

@milvusio reports two Jev passes in their Graph RAG pipeline beat the LLM rerankers they tested on recall while spending less model time.

Two Jev passes helped our Graph RAG pipeline outperform the LLM rerankers we tested, with higher recall and less time spent on model calls. Earlier, we released…

Poster from Sam Gao (@samuel_ys92)

Apps

Sam Gao @samuel_ys92

EN

@samuel_ys92 says one filter can't judge a KOL, an AI researcher, and a founder. The vision he states: diffusion LLMs for context, and @typesafeai's Jev for decisions. DINQ aims to personalize discovery across talent types. He says they are building the best automated talent discovery platform.

One filter can't judge a KOL, an AI researcher and a founder. Our vision: diffusion LLMs for context + @typesafeai's Jev for decisions. DINQ aims to personalize…

Poster from Doug Finke (@dfinke)

Apps

Doug Finke @dfinke

EN

@dfinke built an app that asks Jev questions about service tickets and traces how those answers flow through application rules, so you can watch a judgment become a decision.

What if you could watch an AI judgment become a decision? I built an app that asks #Jev questions about service tickets, then traces how its answers flow throug…

Unofficial developer resource. Not affiliated with TypeSafe AI. These posts belong to their authors. jevnotes.com does not host the videos.