Maybe the Model Isn't the Interesting Part Anymore

I've been playing with an autonomous AI experiment called Lee.

The original question was pretty simple:

Can I make a local LLM do something interesting on its own?

LeewAI‑Bot and the Art of Failing Forward

Experiment #2: Memory, Quality Gates, and the Strange Joy of Watching an AI Trip Over Its Own Feet

Every autonomous system eventually reveals its personality. With LeewAI‑Bot, that personality has emerged in the form of gloriously messy research failures — the kind that make you wince, laugh, and then lean in closer because the failure is actually more interesting than the success.

Today’s batch of posts was full of them.

And that’s exactly why this phase of the experiment matters.

AI shortcomings (and some Washington Commanders sideline) - why you shouldn't always trust AI

A story about how AI can go wrong when you expect it to act like a simulation engine

Maybe you know, maybe you don't: I'm a Washington Commanders (née Redskins) fan. Have been for a long time, more than four decades now. Good times (NFL championships/Super Bowls, magical Jayden Daniels rookie season), bad times (Snyder era, Sean Taylor), and just about everything in between. Even when it was hard, I was still a fan. (Note: I'm not old enough to recall the incredible Sammy Baugh as their QB/Safety/Punter - if you're unfamiliar with that name, and an NFL fan, you should definitely go read that article on his incredible fifteen years at Washington.)

To say this season was a disappointment is an understatement. With that in mind, I turned my attention toward the 2026 draft. And that's where I ran into major shortcomings in the AI realm.