Hey guys, After alot of thinking, Jacob and I have decided to close AI Operators Club. The community closes on October 16th. The reason is simple. We want to put all of our time into actually helping the people who are building, so we're moving everything over to AOC Premium. Running a free community of 16,000 people and a paid one at the same time means neither one gets our full attention, and we'd rather do one thing really well. So you have two options: 1. Save what you need before October 15th. After that the community and the classroom wont be accessible anymore. 2. Keep everything by joining AOC Premium. All the classroom content, the weekly calls with me and Jacob, help when you get stuck and GoHighLevel for free. There's a 7-day free trial, so you can try it without paying anything: https://www.skool.com/aoc-premium/about That being said though. All of our content will, as always, be available on our YouTube channels. Both the existing videos and future videos! No pressure at all, pick whatever makes most sense for you. Thank you so much to everyone who's been part of this. Seeing all the wins posted in here has genuinely been one of the best parts of building this thing. Any questions, drop them below before October 15th 🫡

New video is up. It's the full walkthrough of how Claude Code edits and posts all my short form content. 1. I film a few clips on my phone 2. Claude cuts them in Borumi (retakes and silences out) 3. It adds animations, captions, sound effects and music 4. It makes 3 trial reels in different styles from the same video 5. It writes the captions and schedules everything through Zernio I approve the main video and the trial reels before anything goes out. Resources (skills, setup guide and my exact prompts): https://app.notion.com/p/AI-Short-Form-System-Free-Resources-3ec56c2fc2b381579c3fcae8b346a3d7 Full video is attached below. Let me know if you get stuck setting it up.

We're taking on five service businesses and building our full AI system into them ourselves. Me and Jacob, personally. What's included: 1. A new website built for SEO and GEO, so you show up on Google and in ChatGPT when someone in your area searches for your trade 2. A Google review system that keeps asking your customers for reviews on your behalf 3. Speed to lead, so every enquiry gets an answer before your competitor has even opened his phone Normally that's a setup fee plus $297/month. For these five it's $197/month, no setup fee, and the price is locked for life. There's also a 90 day money back guarantee. If you get no measurable results (no booked jobs, no reviews), you get every cent back. You're probably wondering what the catch is. We want five strong case studies we can document from the first call to the finished build, and we'd rather do that with people from here than with strangers. We'll also only take businesses we genuinely think we can help. If you own a service business, or know someone who does, DM me here or on WhatsApp: +45 52 51 50 61 Full video is attached below 🫡

Anthropic released Claude Opus 5.5 today and I made a quick video on it, but the short version is this. They are claiming it performs at the level of Fable 5.1 on most work, and it costs 40% less to run than Opus 5. So the model most of us actually use day to day just got roughly as smart as their top model and alot cheaper at the same time. The numbers, straight from their announcement page: 1. Input is $4 per million tokens (was $5), output is $20 (was $25). 2. Cache reads went from $0.50 to $0.20. This is the big one. If you use Claude Code, cache reads are a huge part of your bill because it re-reads your project over and over again, and that is why the real world saving is closer to 40% and not 20%. 3. Output is more than 30% faster than Opus 5. 4. There is a new fast mode, up to 2.5x speed, but it costs $8 / $40 so that one is for when you are in a hurry, not your default. 5. On coding it beats GPT-6 Astra. 66.4% vs 57.9% on Terminal-Bench, and basically a tie on FrontierCode (54.4 vs 53.3) at around 20% of the cost. The thing I havent seen many people talk about is the writing. Anthropic basically admitted Claude has been writing in a convoluted way and they trained this one to put the most important information up front and use less jargon. One of their testers said "it writes the way I do". If you have ever had Claude explain a bug in 400 words when you needed one sentence, you know why this matters. Where you get it: it is live in the Claude app on Pro, Max and Team, and they raised the five hour usage limits on those plans at the same time. In Claude Code you just switch it under /model. One line in the fine print you should read before you switch anything: most cybersecurity related requests get re-routed to Opus 4.8. So for that category you are not actually getting the new model. Sonnet 5.5 and Haiku 5.5 are coming in the next few weeks with the same improvements. Full video is attached below if you want the walkthrough with the charts on screen. Source is Anthropic's announcement: https://www.anthropic.com/claude-opus-5-5

My whole feed has been losing it over Jev this week. It's the new model from TypeSafe, built by one of the guys who worked on ChatGPT at OpenAI, and the claim is 200x faster and 400x cheaper than ChatGPT. And to be fair the claim is real. But everyone I've seen is just typing test questions into the box, nobody is actually putting it into a business. So I did. The weird part first. Jev cant write a sentence. It cant chat, it cant even say hello. You give it a situation and a question and it hands you back a decision with a confidence percentage. That's it. Sounds useless until you realise that most work in a business is basically just a yes or no question. Is this a lead? Should I answer this myself? Is this business worth calling? So I had Claude Code build me a small dashboard with five tabs, hooked my actual business up to it, and ran five things I genuinely need done every day: 1. Sort my inbox. 1,000 real DMs from my Instagram and Facebook, three questions each (what is this message, how ready are they to buy, does it need a human). 53 seconds, 3 cents. I then ran the exact same 1,000 DMs through Claude Sonnet, it was 4x slower and 53x the price. 2. Rank 500 local businesses from a lead scraper. Does this business need an AI receptionist, do they have online booking, how professional is the website, do they fit our offer. Done in 14 seconds for about a cent. 3. Every comment on my YouTube channel. Needs a reply, is it a video idea, is it actually a question, is it spam, plus a sentiment score. Around 53 seconds, 3 cents. 4. The one I think is the biggest deal. Jev sits as a gatekeeper in front of an AI auto reply bot. One question: does this need a human? If it scores above 60% (refund, angry customer, something only the owner knows) the bot never sees it and the owner gets pinged instead. Under 60% the bot just answers. Opening hours scored 2%, "I paid last week and nothing happened, I want my money back" scored 98%. Exactly what you want, and the check adds a fraction of a second, not the 3 to 8 seconds a normal model would.