Why I founded an AI company while leading design.

I could have just read about it

There is a version of my career where I stay on the sidelines of this AI moment. I read the reports. I forward the think pieces. I nod along in meetings and repeat the phrases everyone else is repeating. I sound informed. I stay safe.

I did not want that version. After fifteen years of leading and still doing design, I have learned one thing about myself. I do not actually believe something until I have built with it. Reading about a tool tells me what other people think it can do. Building with it tells me the truth.

So I founded an AI company while I was leading design inside a multi-billion-dollar eCommerce org. Not as a hobby. As the only honest way I knew to understand where our craft is going.

You cannot lead a transformation you have only read about

I have led AI-first transformation work inside a large design organization. That work is real and it matters. But here is the uncomfortable thing I noticed about myself and about a lot of leaders around me. We were making big calls about AI based on demos, decks, and other people’s summaries.

That is a shaky place to lead from. When you only consume the story of a technology, you inherit other people’s blind spots. You repeat their optimism and their fear without testing either one. You cannot feel where the tool breaks. And the breaking is where all the real learning lives.

I wanted to feel every sharp edge myself. The only way to do that was to bet my own time on it and build something real that people would actually pay for.

What I built, and why it had to be real

So I founded Story Genie. It is an AI-native company where agents write and illustrate personalized hardcover children’s books, and humans supervise the quality. I built it end to end. The brand. The UX. The web experience. The agentic production pipeline that turns a few details about a child into a finished, printed book.

I made one rule that shaped everything. The customer sees the full book before they pay. Every page. Every illustration. The whole story. If our agents produce something flat or generic or a little bit off, the customer sees it, and we lose the sale. There is nowhere to hide.

That rule was not a marketing decision. It was a forcing function. It meant I could not wave my hands about quality. The product had to be good enough to stand in front of a real parent making a real choice about a gift for their kid. That bar taught me more in a few months than years of reading would have.

The problems nobody warns you about

When you build an AI-native product, the hard parts are not the parts the headlines talk about. The model is not the hard part. The hard part is everything around the model.

The first real problem was trust. A parent is not buying a toy. They are buying something with their child’s name and face and story inside it. That is tender ground. If anything feels careless or fake or cheap, the whole thing collapses. Trust is not a feature you add at the end. It is something you design into every screen, every word, and every moment of the experience.

The second problem was quality control at scale. One good book is easy. A pipeline that produces a good book every single time, for every kind of family and every kind of story, is a different animal. Agents are wonderful until they are confidently wrong. And a confidently wrong illustration in a children’s book is not a funny bug. It is a refund and a broken promise.

The third problem was the one I think about most. Where do you let agents loose, and where do you keep a human in the loop? Get that line wrong in one direction and you drown in manual work and lose the magic of the whole idea. Get it wrong in the other direction and quality slips and trust breaks. There is no clean rulebook for this. You learn it by shipping and watching what actually happens.

Conviction comes from building

Every opinion I now hold about AI and design, I earned by shipping. Not by attending a talk. Not by skimming a research summary. By putting a real product in front of real people and living with the results.

I can tell you where agents shine and where they quietly fall apart, because I have watched both happen in my own pipeline. I can tell you that the interesting design work is not prompt-writing. It is deciding what the human should still own, how to show your work to a customer, and how to make a machine-made thing feel personal and warm instead of mass-produced. I know these things in my hands now, not just in my head.

This is also why I keep insisting on being a leader who still builds. The leaders who stay close to the work are the ones who can actually steer. You cannot delegate your way to understanding. At some point you have to make the thing yourself.

It made me a better leader, not a distracted one

The worry I heard, sometimes from others and sometimes in my own head, was that founding a company would split my focus. That it would pull me away from the teams I lead. The opposite happened.

Running into real problems made me sharper on every team I touch. When someone on a team is wrestling with AI quality, or trust, or where to draw the human line, I am not guessing. I have been in that exact corner at midnight, trying to figure out why the pipeline produced something I would never put my name on. I bring back scars and lessons, not slides.

I am not theorizing about AI-native product design. I am living in it. That changes the kind of questions I ask, the kind of bets I am willing to back, and the speed at which I can tell the difference between a real risk and a fashionable fear. A leader who has shipped is harder to fool and easier to follow.

Designing trust was the hardest and most important part

Of all the problems, the trust problem is the one I am proudest of working through. Making an agentic product that people feel safe handing their child’s name to is its own kind of craft. I wrote about the specifics of designing trust into an agentic product, because I think it is the part most teams underestimate.

The short version is this. People do not trust a machine because you tell them it is smart. They trust it because the experience is honest with them. Show the work. Let them see the result before they commit. Put a human guarantee behind the quality. Be plain about what is happening. Trust is built in a hundred small, deliberate choices, and you only learn which ones matter by getting them wrong first.

Do the thing you tell other people to do

I tell my teams the same things over and over. Build something. Stay curious. Stop waiting for permission. Go find out instead of asking around. Those words are easy to say from a position of authority. They are harder to live.

Founding Story Genie was me taking my own advice. It would have been a little embarrassing to keep telling talented people to be brave and curious while I played it safe and led from a comfortable distance. If I believe building is the way to understand this moment, then I have to be building too. Not someday. Now.

There is a credibility you only get from doing the hard thing yourself. When I ask a team to take a risk with AI, they know I am not asking them to do something I would not do. I already did it. I am still doing it.

What I would tell a design leader still on the fence

A few peers have asked me some version of the same question. They feel the pull to build something with AI, but they have not started. The reasons are always the same, and they are good reasons. No time. A real job that already takes everything. The fear of putting something into the world and watching it flop where people can see. The quiet worry that they are not technical enough to pull it off. I felt all of those. So let me answer them the way I wish someone had answered me.

Start with the time fear, because it is the most honest one. You do not have a free evening sitting around waiting for a side project. Nobody does. But here is what I learned. You do not need to carve out a second career. You need to ship one real thing, end to end, on something you actually care about. One thing. Built all the way through, from the first idea to a real person using it. That single loop teaches you more than fifty articles, and it fits in the cracks of a normal week if you let it be small.

Now the fear of failing in public. This one runs deep for senior people, because we have spent years being the person who knows. Building something new puts you back at the start, where you are clumsy and wrong a lot. I get it. But think about what you are actually risking. A landing page that flops. A tool three friends try and shrug at. That is not a reputation hit. That is tuition. The people whose opinion you care about will respect that you built the thing far more than they would notice that it did not take off.

The impostor feeling is the one I want to push on hardest, because it is the most wrong. You are not behind because you cannot write production code from scratch. The whole point of this moment is that the gap between an idea and a working thing has collapsed. Agents write and build alongside you now. Your real edge is taste, judgment, knowing what good looks like, and knowing when something is quietly broken. Those are exactly the muscles fifteen years of design gave you. You are not unqualified for this. You are unusually well prepared for it.

And the day job, the one that supposedly leaves no room. I will say the thing that surprised me most. Building made me better at the day job, not worse. Every hard problem I hit on my own thing sent me back to work with sharper instincts and real answers instead of borrowed ones. You stop guessing about where AI helps and where it falls apart, because you have felt both with your own hands. That is not a distraction from leadership. That is the part of leadership that is getting hard to fake.

So here is my honest case. It does not need to become a company. It does not need funding or a launch or anyone watching. It needs to be real, and it needs to be finished, because only the finished part teaches you. Pick the smallest true version of something you would be proud of. Build it all the way through. Let it be ugly. Let it be small. You will understand this moment in a way no amount of reading will give you, and you will understand it from the inside, where it counts. Start the one thing. That is the whole advice.

You cannot lead a transformation you have only read about. At some point you have to build the thing, feel every sharp edge, and let the work change your mind.

Where this leaves me

I am a better designer, a better leader, and a more honest voice on AI because I stopped reading about it and started shipping with it. The sharp edges I ran into are the exact things I now help teams navigate, because I navigated them first and have the bruises to prove it.

Our craft is changing fast. I do not think you can understand that change from the outside. You have to put your own time and your own name on the line, build something real, and let it teach you. That is the bet I made, and I would make it again.

If you are wrestling with any of this, where to trust agents, where to keep humans, how to lead a team through it, I am always glad to compare notes. You can reach me here.