My first AI workshop sold out in 4 days. No launch strategy. No ads. Just a text to a few friends and one Instagram post.
I was honestly a little surprised. But when I thought about it, the outcome made sense.
I’d spent about a year before that doing what I call side quests — building AI tools and automations with no agenda other than understanding what was possible. Workflows nobody asked for. Agents that solved problems I’d made up. Experiments that would probably never see a real client.
None of it had an ROI in the moment. That was kind of the point.
When I put together that workshop, I wasn’t starting from scratch. I had a year of experiments behind me. I knew what was possible. I knew where the rough edges were. I knew what would actually blow someone’s mind versus what would fall flat.
The side quests had been the education.
Why “What’s the ROI?” Is the Wrong Question
When most people approach AI, they ask a reasonable question: “What problem does this solve right now?”
It’s a practical filter. It keeps you from wasting time.
But it’s also exactly why most people stay stuck at a beginner level.
The issue is that AI fluency doesn’t work like learning a software tool. There’s no manual. The capabilities change every few months. The way you get good isn’t by studying it — it’s by building things and running into problems and figuring out why they failed.
If every experiment has to justify itself on immediate utility, you’ll only ever try things you already understand. Which means you’ll never build a deep reservoir of experience.
The people I’ve watched get genuinely good at AI fast are all doing some version of the same thing: trying random stuff without pressure. They’re not asking what the ROI is. They’re asking what’s possible.
Gael Breton’s 9 Months
I had a conversation in July 2025 with Gael Breton, who built Authority Hacker into a major SEO and online business platform. He’d been pivoting hard into AI over the previous year.
He said something that stuck with me. He’d spent 9 months in what he called “learning over earning” mode — just experimenting, not trying to monetize any of it.
“I feel like I’m pretty good at this stuff because I’ve literally focused on learning over earning.”
Nine months. No commercial pressure. Just playing.
And he was right that it made him good. Because when he did start building things for clients and for his business, he already knew what worked and what didn’t. He had the reservoir.
What Side Quests Actually Look Like
Side quests don’t have to be big projects. The point is low-stakes experimentation.
Some examples from my own practice:
- I built an AI phone agent that calls my allergy clinic, navigates the automated phone system with timed key presses, and delivers a message on my behalf. Genuinely useful? Not really. But I learned how voice agents work and where they break.
- I organized 15,000 of my own photos using AI — metadata, vision models, folder structures. Something I’d been putting off for two years. Took a weekend. Now I know exactly what photo organization workflows can and can’t do.
- I built an agent that calls and texts me to check in. More of a toy than a tool. But now I understand Lindy’s notification capabilities from the inside.
None of these were client-ready. None of them were revenue-generating. But all of them added to a reservoir I draw from constantly when I’m working on something real.
The Start Small, Iterate Principle
There’s a concept I use in my coaching practice that I call “start small, iterate” — the idea that the best way to build AI fluency is to prove value on a small slice first, then expand. Don’t try to build the whole system at once.
Side quests are the extreme version of this. The slice is tiny. Sometimes it’s just “I wonder if this is possible.”
That question — “I wonder if this is possible?” — turns out to be more valuable than technical knowledge. You don’t need to know Python. You need to be curious and willing to try things that don’t work.
The people who get stuck are waiting for the right project with the right ROI to start building. The people who get good are just building constantly, for themselves, for fun, for the practice.
The Year of Side Quests
I’ve started calling 2025 my Year of Side Quests.
Not because the work was frivolous — a lot of it turned into real tools I use every day.ma But because the mindset shift tters. When I’m experimenting, I’m not measuring. I’m not asking if it’ll work. I’m just building.
The workshop that sold out in 4 days? That came from a year of not worrying about whether things would pay off.
If you want to get good at AI, that’s the path. Not a course. Not a certification. Side quests, consistently, until you have a reservoir you can actually draw from.
Ready to build your own AI tools? Our 4-Day AI Sprint covers the full journey from AI beginner to building working automations — in four days. Over 400 people have gone through it.
Recommended for you
Want the full system? 25X is the flagship productivity system we teach.
Explore 25X →