A few months back I was having lunch with Michael Housman — he does about 50 to 60 AI talks a year, so the guy has seen a lot of formats. We got to talking about what actually works when you’re trying to get someone to understand AI.
His take was blunt: live demos blow people’s minds. Screenshots and videos don’t come close.
And he added something that stuck with me: “10 years ago for us, demos meant something very different. But that still hasn’t changed.”
I’ve been doing AI workshops in Austin since early 2025. Taught hundreds of people. And everything he said matches what I see every single time.
People Don’t Know What’s Possible Until They See It
This is the thing nobody talks about when they’re debating whether to make a video course or teach live.
With AI specifically, the gap between what people imagine is possible and what’s actually possible is enormous. Most people have used ChatGPT to write an email or summarize something. That’s level one. They don’t know levels two or three even exist.
You can describe it. You can post about it. You can make a beautiful 12-minute explainer video. But until someone watches something get built in real time, right in front of them, on a real problem… it doesn’t land.
I call this the AI Fluency Levels gap. Level one is AI-assisted — using AI like a smarter search engine. Level two is building workflows. Level three is deploying actual agents. The jump from level one to two feels impossible until someone sees it happen.
A video can show it. A demo makes them believe it.
The Chef Who Built AI Tools in One Day
At one of my workshops last year, a personal chef showed up. Zero technical background. I think she mentioned she wasn’t even great with spreadsheets.
By end of day, she was building custom AI tools that automated parts of her onboarding workflow for clients. She was asking good questions. She was iterating on her own.
I don’t think she would have gotten there from watching a video course.
Not because the video would have been low quality. But because in-person, she could see something break and watch me fix it. She could ask “what if I change this?” and get an answer in 30 seconds. She had other people in the room doing the same thing, which normalized the confusion.
Watching someone build something live removes a specific kind of fear — the fear that it won’t work for you. When you’re watching a polished video, some part of you thinks “this only works for people who are already good at this.” When you’re in the room and you see the person next to you figure it out, that excuse disappears.
The YouTube Summarizer Moment
I open almost every workshop with the same demo: a YouTube summarizer. You paste a link. In about 20 seconds, you get a structured summary with bullet points, key takeaways, and timestamps.
It’s not the most impressive thing I teach. But the reaction every time is the same.
People get quiet. Then someone says something like “wait, that just… works?”
That’s the moment. That’s what changes someone from curious to interested. And in every workshop, two or three people ask me to set something like it up for them before we’ve even finished the day. That’s not from watching a video about it. That’s from seeing it.
Why Most AI Content Misses
There’s a lot of AI content out there right now. Threads. Tutorial videos. LinkedIn posts with frameworks and diagrams. Some of it is genuinely good.
But most of it has a conversion problem. People read it, think “that’s interesting,” and do nothing.
The reason is that AI is a doing thing, not a knowing thing. You don’t really understand it until you’ve used it on something real. And you don’t believe it applies to your situation until you’ve seen someone in your situation use it.
Screenshots are static. Videos are one-directional. And both require the viewer to mentally bridge from “I see this working for someone else” to “I could do this.”
Live demos collapse that bridge. The viewer is already in the room. The problem is already real. The moment of it working is shared.
What This Means If You Teach Anything
Michael’s been doing this for a decade. The format of “AI” has changed completely since he started. The tools are different. The use cases are different. The audience sophistication is different.
But the thing that actually moves people is the same: watching something get built.
If you teach anything technical or unfamiliar, the fastest path to getting someone to take action is showing them something work in real time.
Workshops. Live demos. Screen shares. Office hours where you build things together.
It’s slower to set up than hitting record. But the conversion rate on belief is not even close.
Try building something live in your next presentation. See what happens to the room.
Thanh teaches AI workshops in Austin and runs the Two Hour Workday program — helping entrepreneurs use AI agents to free up time for the work that actually matters.
