Jev for email — a daily feed

Inbox tools people built with Jev: sorting Gmail, catching spam and phishing, triaging notifications.

121 builds
BRIDGE(JP)@thebridge_jp 2 days ago
Jev email sorting benchmark: 96% accuracy, half cost
【文章を書かないAI「Jev」を試した】 ✅ 応答70〜500ミリ秒で超低コスト ✅ メール仕分け正解率96% ✅ チェック費用は半額以下に #TypeSafe #AIモデル https://t.co/WRfZMBEptM
96% accurate
Email 294 views
Workato Japan 🤖@workato_jp 2 days ago
Slack request triage demo with reactions, replies, and questions
Slackの依頼受付を、内容に応じて自動で振り分ける。 Workatoと外部の判定AI「Jev」を連携したデモです。 ✅ 雑談にはリアクションのみ 👀 具体的な依頼には受付返信 ❓ 情報不足なら確認質問 実際の動きを動画で。仕組みはスレッドで解説します👇 https://t.co/xTKYjlzo2B
Email 111 views
Flowgrammer | AI Automation@Flowgrammers 2 days ago
AP inbox sorting on 100 synthetic invoices
@typesafeai Jev can sort an AP inbox. It cannot do the pay-or-hold maths. A public test with 100 synthetic approvable invoices made that boundary visible; we break down the result and its limits: https://t.co/jZV1K89mqr
Email 30 views
せなお| AIとITを使った仕事術@rutinelabo 2 days ago
Email sorting into reply, payment, and review buckets
📬 𝗝𝗲𝘃 𝗔𝗜 と GPT-6 Astra でメールを仕分けてみた 【返信が要る?支払い?を一括判定】 ・Jevは一瞬で箱に振り分け ・Astraは遅く「自分で見る」に集中 ・判断基準のブレも見えてくる 仕分けだけならLLMはいらないかも。 #JevAI #生成AI https://t.co/siNpr9vSF4
Email 18 views
Dan Shamir@theDushi1 3 days ago
Tiny local QA model trained to answer from text only
קצת דברים רציניים, בשנה האחרונה אני לומד איך לבנות/לאמן/finetune מודלים. איך העולם הזה עובד. מה חדש וכו'. לדעתי זה אחד הskills החשובים בשנים הקרובות. המודל jev הכניס לי משהוא חדש. לקחת דברים מעולם הml הישן ולתת להם טוויסט חדש. אז יצרתי ואימנתי מודל ממש קטן כמה מיליונים של פרמטרים ויצרתי לעצמי את point . מה זה ? מודל קטן מקומי. שהמטרה שלו לשאול שאלות טקסט ואם התשובה שם להחסיר בדיוק את התשובה מהטקסט
Email 4k views
Devan Flaherty@Devanflaherty 3 days ago
DM helper for D&D answers with Jev
Any DND fans around here? Got carried away with Jev and possibly built a bad ass tool for DMs and players who need answers lightning fast, reliably, with no hallucinations or creative input. And it does a little more too :). Looking for some testers! DM If interested. https://t.co/jCTVmfnEOy
Email 440 views
ziya@ziyacivan 3 days ago
Gmail filters in plain English with Jev
@EGafni Here you go: Gmail filters written in plain English, judged by Jev 👇 https://t.co/cg2oymmDZG One Apps Script file, no server. Each new email → one Jev call with a yes/no question per rule → label/archive. Setup takes ~2 min.
Email 138 views
Juana IA@jvsalomz 3 days ago
Email triage with Jev, 100 emails in 1.4 seconds
Pega esto en tu Claude Opus 5.5 + Jev y mira la diferencia. Jev clasifica cada tarea antes de gastar tokens. Opus solo razona cuando hace falta. Resultado: 100 emails triageados en 1.4s, contexto 1M→86k tokens en 1s, safety gate 18x más rápido. LLMs piensan. Jev decide. https://t.co/9LEz5wHFSi
100/s18× faster
Email 26 views
コトノハ*ツムギ|writing/design/AI/Canva for NPO@cotonohatsumugi 4 days ago
Disaster volunteer center email triage demo with Jev, 160+ emails in 2s
Claude Code✖️Jevで、災害ボランティアセンターのメール判別デモを作ってみた。こんな一次トリアージができたら、災害ボランティアセンターもすごく助かるんじゃないかと。160件以上のメール、2秒台で判定してくれてる。災害時にVCを担う社会福祉協議会の皆さんの参考になるかな。 https://t.co/EVB8HvNeYG
Email 60 views
Omar D.@oadiazp 4 days ago
Gmail classifier reranked 408 unclassified emails to zero backlog
I replaced my local Kev Gmail classifier with Jev via OpenRouter. Jev re-evaluated 408 messages left as “Not classified” — and the backlog reached zero. The catch: more decisions isn't the same as more accurate decisions. https://t.co/Vx2BguOihd
Email 20 views
Vxrcel@kleen_pulse 4 days ago
Raven mail client with reply briefing and mail search
I built Raven, a self-hosted mail client that briefs you before you read. It works with Gmail, Outlook, iCloud, Fastmail or any IMAP mailbox. The home screen is built from your last 30 days of mail: replies you owe, people who owe you, and what can wait. Press Ctrl+K and ask "who's waiting on my reply?" The answer shows the emails behind it and exactly what was searched. It runs on Jev, a decision
Email 7 views
Karol K@iamkarolk 5 days ago
Email classifier for senders and subjects with Jev
Jev adds classification on top, guessing what each email most likely is. Results are okay given I only send it senders and subjects, not entire messages If you're struggling to free up Gmail space too, check it out 👇 https://t.co/uYZtIoKa5A
Email 26 views
maestro@maestrooth 6 days ago
Inbox sorter with Jev for overnight triage
JEV + GROK BOT is insane... my inbox now sorts itself overnight Grok Bot reads a shared mailbox on its own cloud computer while i sleep. Jev asks four questions about every message in one call By morning there's a short list and drafts waiting, and nothing has been sent. [copy this into Grok Bot:] "Every night at 1am, open the shared mailbox and go through everything that arrived since the last ru
Email 2k views
GeekNews@GeekNewsHada 6 days ago
25-line local phishing classifier with Qwen3-0.6B
Python 25줄로 구현한 Jev Qwen3-0.6B에 질문과 A/B/C 선택지를 주고, 답 대신 선택지 토큰 점수를 확률로 바꾸는 예제임 수상한 급여 이메일을 피싱 88.5%, 스팸 8.4%, 정상 3.1%로 분류함 API 없이 로컬에서 처리하며 합성 데이터 학습과 확률 보정은 빠진 패러디임 https://t.co/R1hMPWqPCy
Email 1k views
池田龍太|高卒司法書士イケダくん@souzou_office 6 days ago
NARAU Gmail triage tool using 35 features and learned weights
Jevの使いどころが少し見えてきた。最終判断をさせるより、判断材料を作らせる方が面白い。 Jevは文章を書かない判断専用のモデルで、「このメールは返信が必要?」のような質問に0〜1の確率だけを返す。自作のNARAU(Gmailを「見る/見なくていい」に自動で振り分けるツール)で使っているけど、「このメール見るべき?」とは聞いていない。 聞いているのは、「返信が必要か」「作業が必要か」「営業・メルマガか」「自分への依頼か」「期限があるか」「エラーや異常か」といった観点ごとの質問。こういう特徴を27個作って、Gmailから機械的に取れる8項目と合わせた35個の数値に、自分の過去の判定(見た/見なかった)を突き合わせて、どの特徴をどれだけ重視するかの重みを学習させている。 Jevが測るのは「このメールがどんな性質か」だけで、「その性質を自分がどれだけ重視するか」は重みの方で決まる。「広告っぽい」
Email 667 views
Omid Sayfun@OmiidSayfun 6 days ago
Chrome extension that labels important unread emails with Jev
Couldn't find my important unread emails, so I built a Chrome extension. Jev finds them and labels them by what they're about. You can check it out here: https://t.co/Ktaa3eD4Z4 https://t.co/e0csLl15YY
Email 503 views
PromptDesk AI@PromptDeskAI 6 days ago
Pipeline for comments and replies filtered by Jev, 79 ms
It's 4:28 a.m. and I'm still up, because after nine evenings of trying to break Jev, it finally held. Tonight it's in front of Grok 4.7 for real. Every LinkedIn comment, YouTube reply and Telegram message in my pipeline hits Jev first. Six yes/no questions, under half a second, five cents per thousand. Grok only ever sees the ones that pass, and drafts the first message. I send. Deciding is 79 tim
79× faster
Email 38 views
Fermex95@fermex95 6 days ago
Local email classifier deleted 62,025 mails and freed 12GB
Hotmail full at ~70,000 emails. Instead of paying for more storage, I built a local Next.js + SQLite app that uses Jev to classify every email, with rules protecting anything important. 62,025 deleted. 12GB+ freed. $3.64 in tokens. Full write-up: https://t.co/pt0dRFivnn
62,025 items$3.64
Email 30 views
🩶🤍🖤🤎❤️🧡💛💚💙💜@deslopper 6 days ago
Email cleanup tool with Jev metadata screening and model review
写了个工具自动清理过期邮件、广告邮件。 jev先用元数据初筛,不确定再带正文送入chat模型复核。 很省钱。之后考虑用类似的本地模型替换jev。 https://t.co/rIKzsnKGIl
Email 29 views
Kojo Jeffery@openclonne 6 days ago
Personal email classifier and proactive email workflow with Jev
Sent a whatsapp voice note to Warden (My Personal Agent, managing life and homelabbing) - to build an email classifier and a simple way to be proactive with my emails with analytics. Classified by @typesafeai Jev-latest, framework should be laravel, cordinate through @papercliping by reporting to the CEO and a daily cron to run the announcement skill using edge-tts. Goodness! all done in 30min - U
Email 14 views
Webenable@Webenable_nl 6 days ago
Email classifier workflow using Jev for daily work
Toen ik hoorde dat Jev heel geschikt was om e-mails te classificeren, ging ik natuurlijk meteen aan de slag. Sindsdien is mijn werk een stuk leuker geworden! https://t.co/xT01nqOTC6 https://t.co/j0ccI6L4na
Email 13 views
Vxrcel@kleen_pulse 6 days ago
Email classification and management with Jev and Claude Opus
JEV + Claude Opus 5.5 = super powered email classification and management https://t.co/sTMt3YreHj
Email 10 views
Jasper Li@Jasperli0122 Sep 23
Email reading triage, 27 emails in 2.68 seconds
Jev + Monid is insane for reading emails. 27 real emails. One call. It caught 2 phishing attempts, flagged 12 that needed replies, and ignored the other 15. 2.68 seconds. $0.000476. https://t.co/1Np5QUW7kn
27/s$0.00052.68 s
Email 634 views
ようへい@表現者の才能を事業化する中の人@40jobseeking Sep 23
Email triage site using Jev for real sorting
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。 https://t.co/6h0Isq87cl
Email 76 views
Miguel Garcia@MikezGarcia Sep 23
Built a customizable email filter that catches Gmail spam
Jev helps me keep my inbox clean at a low cost 📬 I built a customizable email filter that catches spam Gmail misses, learns over time, and adapts to what I actually want to receive. https://t.co/JHDNjYmcXR #Jev #TypeSafe #BuildInPublic
Email 44 views
Cogito@0xChainThought Sep 23
2,000 email classification test, 62.6% to 95.0% with enum split
Same model, same 2,000 emails. Jev scored 62.6% on day one and 95.0% once the question got split into five. Nobody retrained it. Someone rewrote the list. We wrote about the enum gate before: a decider that can only choose from allowed answers can't hallucinate. This test adds the other half. It can still pick wrong, and the fix lives in how you cut the question. The bill follows the cut. Haiku wi
62.6% accurate95% accurate$930.75
Email 42 views
Git_Shark@LFrefman Sep 23
25-line phishing classifier using local Qwen3 logits
觉得Jev吹得太玄看不懂?这篇25行Python的恶搞复刻可以看看。 NobodyWho用本地Qwen3-0.6B演示了Jev的本质:给邮件出三选项,直接拿A/B/C三个token的logits算概率,示例把假冒Payroll页判为Phishing 0.885。 1️⃣ 本地加载Qwen3-0.6B-Q8_0.gguf,开logits_all就能玩,不用调任何API。 2️⃣ prompt就是单选题模板:Email内容加A. Legitimate / B. Spam / C. Phishing。 3️⃣ 只取最后一步对A、B、C的logits,做softmax就得到校准过的概率分布。 4️⃣ 全程讽刺拉满:不需要System One、合成数据和RLCD,数据也不外发。 一句话:Jev没那么神,就是本地小模型的单选题分类器。 https://t.co/z8Hb6A0fwI https://t
Email 34 views
せきのです@sekiemon_gb350 Sep 23
Qiita article on using Jev to speed up answers
https://t.co/RImOgA9IXt 書きまして。 全文出てから推測ではなく、途中の推測(jev)で返答書かせる事でスピードアップを図るというもの
Email 22 views
Leaf Yeah!@leaf_sanren Sep 23
Chat copilot that judges intent, risk, and reply priority
啊啊啊啊啊🎉🎉!JEV 黑客松已经出来啦!!! 同时,我也见到了把 Jev 用得最对的一个项目!! Jev 这模型最大的特点: 只会判断,一个字都写不了。所以用它的关键,就是别让它干写字的活。 这个开源的「聊天副驾」就很聪明: 你在微信里收到一句"在吗"或者"方案再想想",它先让 Jev 一秒判出对方真实意图、危险等级 1–9、该马上回还是先晾着; 写回复这活交给另一个会写字的模型起草 3 条,最后再让 Jev 排个序。 判断归判断模型,写字归写字模型。 且它最好的一点:它永远不自动发送。 搞得我都想去闲🐟搞几台吹灰的安卓机啦! 期待后续也能拿下 iOS、macOS、鸿蒙系统! https://t.co/6xTAqkS9zF
Email 19 views
Medical Boulevard@JoinMedicalBlvd Sep 23
Patient message inbox triage into scheduling, billing, and review
The patient-message inbox is full on Monday morning. What should a person see first? An LLM reads each message and pulls out the request. Jev sorts it into set queues: scheduling, billing, records, clinical review. Anything clinical or unclear goes straight to staff. Jev is early access, not clinically validated.
Email 11 views
Vxrcel@kleen_pulse Sep 23
Auto-classification of mail with JEV
Auto classifying mails with JEV https://t.co/LXMCimjg2b
Email 6 views
Atlas Wegman@iiAtlas Sep 23
Mail spam filter that moves low-value email and tracks orders
@OpenRouter @jjacky Absolutely loving @typesafeai Jev on OpenRouter! Built my dream mail spam filter. Runs hourly, moves "Low Value" emails out of my inbox, and tracks orders in Parcel automatically. You'd be surprised how much more sane you can make email by simply classifying "low value"! @jjacky https://t.co/aHdSmWUcdh
Email 6 views
G.O2@_pago2_ Sep 23
Sorted HR inquiry emails with Jev
"判断するだけのAI"Jevで、人事の問い合わせメールを仕分けてみた!|G.O2 @_pago2_ https://t.co/1QvDB9AdDk
Email 4 views
【公式】Jinba | AIエージェント開発@jinbaflow_JP Sep 22
Slack request router that triggers Jinba Flow with Jev
JevでSlackのメンションやめてみた。 普通に書き込むだけ。判定AI「Jev」が発言を読み分けて、雑談はスルー、仕事の依頼だけ拾いJinba Flowの必要なフローを自動で起動。経費ルール検索も資料生成もその場で完結。 声をかけずとも、必要なものが向こうからやって来ます。 https://t.co/hEQXxsNHtl
Email 20k views
池田龍太|高卒司法書士イケダくん@souzou_office Sep 22
NARAU Gmail sorter that learns what to read from your feedback
JEVを使って、自分の「見る/見なくてよい」を学習し、Gmailのメールを自動で仕分けるソフト「NARAU」を作りました。自分が回答・訂正した結果が、次のメールの振り分けに反映されます。 仕組みは、1通のメールを9つの観点から判定するところから始まります。「返信が必要か」「具体的な作業を求めているか」「広告か」などをJEVがそれぞれ数値にして、そのメールの特徴として保存します。 自分がすることは、そのメールに「見る/見なくてよい」と答えるだけ。9つの数値と回答をセットで蓄積し、「こういう数値の組み合わせなら、自分は見る/見ない」という傾向を学習します。 料理にたとえると、すでに「辛い」「肉が中心」「揚げ物」といった特徴が分かっていて、それに「食べたい/食べたくない」と答えていく感じです。回答がたまると、「この人は辛い料理を好むけれど、揚げ物はそれほどでもない」といった傾向を学び、次のおすす
Email 718 views
davepoon@davepoon Sep 22
Email generation benchmark: 46.3s vs 2.0s on Jev
Both emails building at the speed they really built. At 4.2s Jev has masthead, greeting, headline, intro, callout and footer down. The writing path has not placed a block. Medians, 3 runs: ours 46.3s Jev at once 2.0s Jev slot by slot 4.7s Yes, 46s is ours. https://t.co/elaQxVOuoZ
Email 373 views
85 more below

Submit a build

Paste the link to your post on X about something you built with Jev. We check the link right away. The next update, within about 4 hours, adds the card if the post shows your build.