I love AI. I use it in almost every aspect of my business, from coding and analysis to strategizing and research. But the more I use it, the more convinced I am that many people are starting in the wrong place. They start with AI. They open ChatGPT or Claude and ask, “What should I create?” “Give me some ideas.” “Build me a game.” “Create a framework.”
I tend to approach it differently. I usually already know what I want to accomplish and I use AI to remove whatever is standing in the way.
In an earlier article, I talked about how I needed to cut my 60-minute Personality Poker speech down to a 12-minute experience for a showcase I was doing. I could have asked AI what to cut, but that would have meant letting AI decide what a good Stephen Shapiro speech should look like. I know what works. I know what’s important. I can remember audience reactions and questions they ask after an event.
So instead of asking AI to edit, I gave it a time-coded transcript from the full-length 60-minute version that I knew worked well. That became the baseline. Then I rehearsed the new version over and over, feeding each time-coded transcript (an SRT file) into AI and asking it to compare my performance against the baseline. Where was I spending too much time? Where was I speaking faster? What sections had expanded? What could I tighten without changing the experience? Where was I stepping on the punchlines?
On the day of the event, I delivered the experience in 11 minutes and 55 seconds. AI didn’t create the speech. It coached me on something I had already created. That’s an important distinction.
Another example started with an audience member’s question after I gave a Personality Poker speech. He asked me if I had looked at the best way to configure teams.
I’ve been researching this topic for two decades. I’ve looked at when teams outperform individuals, when individuals outperform teams, and how the complexity and ambiguity of work influences how we configure the people doing it. I had already written numerous articles on the research and had several chapters in You’re Not Playing with a Full Deck about how the nature of the work influences team configuration.
While flying back home, I started playing with the idea of how I could “automate” the decision-making process. Sitting in the airport lounge, I laid out the strategy. What did I want the system to do? What were the inputs? What were the outputs? What was the process?
On the plane, we had Wi-Fi, so I started developing the app with Claude Code while flying at 30,000 feet. By the time I landed, I had a working interactive application that let people explore the concepts and apply them to their own work. Ten years ago, that probably would have required finding a developer, explaining what I wanted, getting a proposal, spending thousands of dollars, waiting weeks or months, testing it, and making revisions. More likely, I simply wouldn’t have built it.
Now, instead of binge-watching Netflix on the plane, I build a fully functioning app.
The point is that the idea wasn’t created by AI. The research wasn’t created by AI. The intellectual property wasn’t created by AI. AI simply allowed me to go from idea to implementation in one afternoon.
My favorite example goes back even further. More than a decade ago, I envisioned taking my Personality Poker experience and creating a system that could “read” people’s hands. The idea was simple: take a photo of your five Personality Poker cards, upload the image, and then interpret what it meant. Who you are. Who you are not. Who you need. Your blind spots. And more.
More importantly, I wanted to aggregate all of that information in real time so I could provide insights into the composition of the teams and organization while on the stage, and then provide the data to the client after the event.
Twelve years ago, I spoke with a developer about this idea. Some of what I wanted wasn’t technically practical at the time. Other parts would have cost well into six figures. So the idea sat there. The idea didn’t suddenly become better twelve years later. My ability to get it done did.
Today, I’ve built what I call the “Full Deck IQ” system. It does all of the things I imagined years ago and quite a bit more. During a live event, hundreds of people can participate, receive individualized insights, and collectively create organizational data that I can use in real time.
Again, AI didn’t give me the idea. It made the idea possible.
These experiences and others have changed the way I think about AI.
I see three particularly powerful roles for it.
- It can be a coach, helping us improve things we already do. For example, it helped me improve the timing and pacing of my speech.
- It can be a builder, helping us turn our ideas, expertise, and intellectual property into things that previously required technical skills, significant money, or large teams. The configuration software I built on the plane is one example.
- It can also be an amplifier, extending the value of what we do beyond a single interaction, meeting, workshop, or speech. My Full Deck IQ system does this well and has also enabled me to find new revenue opportunities.
But there is another side to this. As AI makes creation easier, the ability to create something becomes less differentiating. Judgment becomes more important. Anyone can generate an app. Anyone can create an assessment. Anyone can produce a framework that looks polished and impressive. That doesn’t mean it’s good.
You still need to know what problem is worth solving. You need experience, expertise, and a point of view. You need to know what you are trying to accomplish before you start building. And perhaps most importantly, you need the judgment to recognize when something AI produces looks impressive but isn’t actually useful. I’ve seen plenty of AI-generated output that looks fantastic on the surface but falls apart once you start examining what’s underneath.
That’s why I don’t think the most interesting question is, “What can AI create for me?” I think a much more interesting question is, “What have I always wanted to create that I couldn’t?”
Maybe it was too expensive. Maybe it required technical skills you didn’t have. Maybe it was too complicated or too time-consuming. Maybe you had the idea years ago but couldn’t justify the resources required to make it real.
Go back and look at those ideas again.
AI may not give you your next great idea. It may make one of your old great ideas possible.