What You’ll Learn James has 750 YouTube videos and, at the time of this session, exactly one public short. This is how he built a shorts pipeline he owns outright — no per-clip credits, no third-party service holding the output. The Problem With Clip Tools He’s clear that tools like Opus Clip do the cutting…
What You’ll Learn James built a pipeline that turns his videos into branded, captioned shorts with no editor and no timeline. It ran. It produced ten clips. They were unusable. This is the most useful kind of session — a failure narrated honestly — and the lesson generalises far past video. What Actually Worked Almost…
Forget six-month launch plans. The new way to make a digital product: capture something you already do that works, build it with your AI chief of staff in about an hour, and save a reusable playbook so the next one is one sentence.
Live, James builds a sellable digital product — the Prep Agents Team plugin — in about 90 minutes using Fable 5 and his orchestrator Dean, while capturing a reusable playbook so the next one takes a single conversation. The lesson: productize the valuable task you already repeat.
Live on Fable 5, James launches a $97 product with a single prompt — his orchestrator Dean builds the course, emails, social posts, community campaign, and shopping cart in about 90 minutes. Inside: the three stages of shipping a product and the memory-plus-wiki system that makes it work.
James runs Sonnet and Mythos 5 head-to-head on the same text-heavy lesson. Sonnet returns a wall of styled text (and grabs the wrong lesson); Mythos diagnoses the lesson, then builds an interactive visual style picker in a single prompt and embeds it in the course.
James builds a full mini-course in about 15 minutes with Dean — written, published to the LMS, and ready to sell — to make a bigger point: content courses aren’t a saleable business anymore because anyone can generate one. Inside: the three parts of an AI operating system and where courses still belong.
In a live demo, James runs CourseLab to rescue a static lesson: it reads the content, diagnoses the best interactive asset, builds a pricing simulator, quality-checks its own screenshots, and publishes it — even working around a WordFence block it had learned about from a prior run.
James shows how CourseLab, an AI agent team, rescues old static courses and rebuilds them into interactive, visual lessons students actually finish — complete with a self-QA step and a downloadable skill for learners.
Anthropic’s Mythos/Fable 5 isn’t about benchmarks. For educators, the real story is a visual engine that reads your images, builds living visual assets as code, runs hour-long workflows, and remembers more. Here’s what changes for your teaching — and the one thing to do this week.
An Anthropic engineer just admitted on a podcast that his markdown plans got so long he stopped reading them. If he’s not reading them, you’re not either. Here’s the one-prompt fix — switch your Claude Code plans from .md to .html — and the 5 patterns to try tomorrow.
What You’ll Learn James has 684 YouTube videos. If you asked him what’s in any of them, he couldn’t tell you. That’s not a storage problem — it’s a memory problem. And AI tools like Claude have the same problem: every new session starts from zero. In this video, James walks through how he solved…