At Morningside, we don't sell AI. Strange thing to say when we've worked with Aston Martin, NBA teams and publicly traded companies implementing it. But businesses don't care about the latest AI tool you found. They care about ROI. Over the last few weeks I've shown you different ways to get clients - including walking into a business and booking a meeting in under 20 minutes. Which creates the next question: What do you actually sell once you get the meeting? Next week we're pulling back the curtain on what we're seeing AI agencies sell in 2026. Completely free. Four years of running Morningside, plus what we're seeing across 7,150+ agency owners inside the Accelerator. What businesses are buying. What's working right now. Where we believe the opportunity goes next, based on the data we have. We're doing things different for this event: 1. This is the only post I'm making about it. This will be the last post about it. 2. I'm capping the session at 1,000 people. We usually get 6,000+ registrations. If you want a spot, move fast. Registrations are now open. 6:00pm EDT, Wednesday 23rd September. GRAB YOUR SPOT → https://live.aaaaccelerator.com/us First 1,000 in, then we close it.

GPT-6 Astra and Claude Fable 5.1 came out within a few days of each other, and the "which one is better" posts started almost immediately. I collected the main numbers in the chart above. Honestly, they're closer than I expected. They score the same on the Artificial Analysis Intelligence Index, cost the same per token, and both handle a 1M-token context. Astra gets through tasks for a lot less money and seems stronger at terminal work and automation. Fable 5.1 looks better on long-running projects and big codebases. The more I looked at it, the less it felt like a competition. Their strengths don't overlap much, so I started wondering if the smarter move is to just use both. Let Astra build something, then have Fable 5.1 review it. Or the other way around. But I'm not sure that works out in practice. Two models means two sets of prompts to maintain, two APIs to deal with, and two bills. And when something goes wrong, it's harder to tell where the problem came from. So I'm curious what you think. If you were starting a serious project today, would you pick one model and really get to know it, or use both and let them cover each other's weak spots? If you've already tried either of them, or both, I'd love to hear how it went.

Hello, As you probably know, a lot of US pros are taking on more work and delegating smartly to boost their earnings. I've been on the receiving end of the delegation and seen firsthand how well it works. Now, with AI, you can take on more projects while I handle the workload efficiently. I manage most of the project lifecycle (incl. winning a contract), and I just need your help at a few key moments...nothing too technical or difficult. Let's talk.

I'm an engineer by trade and want to make my own products. Building is easy due to my background and AI, but getting paying users is hard. I've seen people who make not-so-good products, or who aren't very technical, still get their $10k consulting deals or paying customers. I had a reflection that I need to have something to show online, so I run into rabbit hole making videos. The first classroom really opened m eyes, and I'm so grateful the owner shaped my mindset at the first classrom. Glad to be here!

I have finished a workflow automation project for a growing agency. For the task, I evaluated two models, GPT-6-Astra and Claude-Fable-5.1 – both have performed satisfactorily, but the latter has produced lower-quality results. GPT-6-Astra was able to achieve excellent results in a shorter number of steps – specifically, a complex instruction required four steps to execute. In comparison, Fable-5.1 took five steps and produced inferior results, consuming 1.5x more tokens and taking twice as long. I have previously favored Claude over GPT, but this task has made me seriously consider which model to choose for planning tasks. Would you like to vote?