Frameworks I actually use, opinions I'll defend, and the occasional case for not using AI at all.
Kicking off one agent that spins up 45 more isn't running a fleet — you typed one thing. The hard part was never producing more; the judgment layer doesn't scale, because the judgment layer is a guy.
Fine-tuning, RAG, prompting — a plain-English tour of what each level of AI training actually costs you and when it's worth it.
Why the thing that writes you a sonnet can't reliably count the letters in it — and what that means for what you ship.
Ideas are cheap and execution is expensive. Here's how to find the opportunities worth spending execution on.
A compressed, opinionated path from a vague idea to something a real person can click on by Friday.
A fast, unglamorous viability check you can run before you fall any further in love with your idea.
I finished Transformed and The Mom Test in the same week. One made the top shelf. The other is 340 pages where only 40% is new — here's the scorecard against its own three big goals.
Everyone claims GenAI experience; most have just used ChatGPT. The Kaggle-to-DeepLearning.ai path I hand anyone who asks how to genuinely learn this stuff.
126 pages, zero filler, and it shows you what good and bad customer conversations actually look like instead of staying theoretical. Welcome to the Hall of Fame shelf.
You proved the ROI and two weeks later nobody remembered the deck. 10x ideas are a dime a dozen — memorable ones aren't. The first three principles of sticky ideas.
I turned my resume into an IBM Watson chatbot: the average resume gets six seconds, I wanted fifteen minutes. The original Brandy — she's come a long way since.
WAIT — Why Am I Talking? Three habits I took from Kevin Sharer's piece on going from know-it-all bully to someone who actually listens.