I just found an open-source model called Quyet. It doesn't generate text. It reads a message and picks an option, with a % confidence for each. Example: a customer writes "Close my card, I lost it yesterday." → Wants to cancel: ✅ → Urgent? Probably yes → Mood: annoyed Imagine that on your incoming leads: hot or cold, price question or legal question, urgent or not. The catch: I haven't tested it yet, and I have no idea how it handles real-world Vietnamese messages. 🤔 Has anyone tried it? Would you trust a model like this to sort your leads? github.com/ncchinh/quyet

Hi everyone! Someone know or have any automation to save time doing Monthly SEO Reports? Thanks in advance :)

hey everyone 👋 i made this week's pro update public so you can see what we're building. - everything is moving into one markdown workspace (AI Marketing Hub Home). the old brains get archived after porting - the agent fleet: one view to watch every agent run, per client - Jev SEO: testing a tiny, fast model for keyword and SERP sorting. my early guess is around 3x fewer tokens. not measured yet video: https://www.youtube.com/watch?v=-etyKAf1ncI free takeaway you can use today: run one SEO audit, write two posts, then ask your agent to schedule both every week. Home and the fleet live in pro. staying in the free hub is 100% fine, and everything public stays public 🙌

Ran Daniel's free Jev SEO skill on odgtravel.com (our family & group tour company in northern Vietnam). Sharing the results in case it helps anyone with a small business site. 📊 The numbers 85/100, grade B. 60 URLs crawled, 59 pages analysed, 752 Jev judgments, total Jev cost $0.0124. 21 actions found, 4 critical or high. Partial audit: I had no PageSpeed key, so performance only uses crawl timings. ✅ What's working Security and trust 100. On-page, structured data and performance 92. All 10 AI crawlers it checked (GPTBot, ClaudeBot, PerplexityBot, Google-Extended...) are allowed in. ⚠️ What hurts Content quality is the weak spot at 70. Jev flagged 9 important pages (destination pages, a Ha Long cruise page, blog, combo tours) as not really answering what visitors search for, averaging 0.28 on a 0 to 1 helpfulness scale. Basically: lists of tours, not enough real writing. It also found some pages competing for the same topic, and slow server responses (some pages over 7 seconds at crawl time). 🧠 Gotcha for Cloudflare users 59 internal links pointed to a 404 at /cdn-cgi/l/email-protection. That's Cloudflare's email obfuscation. Real visitors usually see the email fine, but crawlers that don't run JS see a broken link. 💡 My takeaway The technical side was mostly fine. The real gap is content that actually helps a traveller decide. AI can't fake that part, it needs real trip details and real experience. Question: for those running local or service businesses, which scored lowest for you: content, performance or crawl? Curious if it's the same pattern.

Jev ~ duno, but I just joined the waitlist for early access etc https://typesafe.ai/ ~