Something shifted in how I started sharing my work.
I started live streaming. From my terminal. From Teams sessions. Showing exactly what Claude Code was doing in real time.
The response was immediate and honestly surprising. People didn't just want to see the output — the finished feature or the refactored code. They wanted to see the process. How did I phrase the initial prompt? What happened when the AI went down a wrong path? How did I course-correct? What did the recovery look like?
They wanted to see the messy, iterative, sometimes frustrating reality of AI-assisted development. Not the highlight reel. The actual work.
I started recording more carefully. Narrating what I was doing. Explaining my reasoning. Showing the AI's mistakes and my responses to them. Keeping the audio quality good even when I was doing it from my car on the way to the summer house (okay, that one was imperfect, but the point came through).
What I realized while editing these sessions is that I've been accumulating something I can't easily transfer through written posts: tacit knowledge. The feel of working with AI. The intuition about when to break a task down further. The sense for what's going wrong by reading the error messages. The way you phrase a follow-up prompt based on what the AI just produced.
It's the same thing a chef learns from years of cooking but can't fully teach in a recipe. It's the same thing a musician learns from playing but can't fully notate. It's embodied knowledge.
Video is the medium for tacit knowledge. Text is for explicit knowledge.
I'm not abandoning writing — this blog series exists because I believe in the explicit layer, the "here's what I learned and why." But the streaming sessions are becoming the other half of the story. They're where I show instead of tell.
The side effect is that people started asking more sophisticated questions. They'd watched me work and now they understood the tools at a different level. They could see the patterns. They could try things themselves and come back with variations.
That's learning transfer. That's community.
I don't know how to scale this beyond what's sustainable for me to record, but it's clearly more valuable than I initially thought. The demos matter. The polished output matters. But the how matters more.