Harley and I were sitting on the living room floor this winter surrounded by soccer card packs. Topps Chrome, Donruss, Prizm. He'd rip one open, pull out something shiny, and immediately want to know: "Dad, is this one worth anything?" I had no good answer for him. So I'd pull out my phone, fumble through Beckett or PSA or eBay, and five minutes later I still wasn't sure.

Soccer cards exist on the big sites. Beckett has them. SportsCardPro has them. eBay has plenty of listings. But the experience is terrible. These platforms were built for Adults. A Young User is an afterthought. Try finding the value of a specific Topps Chrome UEFA Club Competitions parallel and you'll spend more time searching than the card is worth. The data is scattered, the interfaces are clunky, and nothing is built for the way soccer collectors actually think about their cards.

Here's the thing. The soccer card market is growing fast. The World Cup is coming to the US this summer and interest is surging. New collectors are entering the hobby every week, and they're all hitting the same wall Harley and I hit on that living room floor. I'm a product manager. I see gaps like this for a living. The difference this time is that instead of filing it away as "someone should build that," I built it. In about a month. Without writing a single line of code. With my 8-year-old as the product manager and QA team. The result is pitchpackcards.com.

The Barrier Is Gone

I don't write code. I've never written code professionally. I built PitchPack using Claude Code and Codex, talking to AI the same way I talk to engineers on my team. I describe what I want. I explain why it matters. I provide context about the user, the problem, and the constraints. Then the AI builds it.

This isn't "no-code" in the way people usually mean it. There's no drag-and-drop builder. No Squarespace template. PitchPack is a real web application with AI-powered card scanning, crowdsourced pricing, a community database, and a search architecture that handles thousands of cards across dozens of sets. I speak to AI like a product manager because that's what I am. And it turns out that's exactly the skill that matters.

I build plans in Claude Code before I implement anything. I think through data structures, user flows, edge cases. I map out how the search should work before I ask the AI to build it. The planning and product thinking IS the skill. Code is just the output. A month from first idea to live product, and the hardest parts were never about code. They were about planning. Database structure. Search architecture. Scraping set reviews, player data, parallels, pull rates. Getting the foundational setup right mattered enormously because redoing data work at scale is expensive, even when AI is writing the code for you.

What started small grew into something much bigger than I expected. The first version was a simple card lookup tool. Now it has AI scanning, set reviews with pull rate analysis, player profiles, parallel guides, and community pricing. Each feature started because Harley or I hit a wall while using it. That's how good products grow. Not from a roadmap document, but from using the thing you built and noticing what's missing.

Product sense was the skill that mattered most. Not engineering chops. Understanding what to build, in what order, and why. Knowing which corners you can cut early and which foundations you need to get right from day one. That's product management. And it's the thing AI can't do for you.

The Best PM I've Ever Worked With

Harley sits next to me while I build. He uses the site constantly. He finds cards, scans them, looks up values, browses sets. And somewhere along the way, he started making product decisions that were genuinely better than what I would have come up with on my own.

The card scanning feature is a good example. Early on, the AI identification wasn't perfect at distinguishing between parallels. A base card and a silver ice parallel can look almost identical in a photo. I was deep into improving the computer vision when Harley said, "Why can't I just type in the card number and tell it what set it's from? I already know that." He was right. Instead of making the AI do everything, let the user narrow the problem. He already knew the card number and the set. He just needed help identifying the specific parallel. That's sophisticated product thinking: don't force AI to solve the whole problem when the user can eliminate half the variables with two inputs.

His feature ideas kept coming. He wanted a collection feature so he could track what he had. A trade feature so he could swap with friends. A "top players" page because he liked browsing the expensive cards and seeing what was out there. He wanted parallel images shown during the identify workflow and on card pages so he could visually compare his card against known parallels. Every one of these came from using the product himself and hitting a moment where he wanted something that wasn't there yet.

Then there was the ROI calculator. Harley thought every set was amazing. He wanted to open everything. As a parent, I didn't want to spend unlimited money on card packs. So we needed a way to figure out which packs were actually worth buying. Which sets had the best odds of pulling something valuable relative to the cost of the box? That feature exists now because an 8-year-old's enthusiasm met a parent's budget constraints. He even insisted on confetti animations when you pull a rare card. Delight as a feature, not an afterthought. He was right about that too.

Harley's value isn't despite being 8. It's because he's 8. He's the actual user. He lives this problem every single day. He doesn't overthink solutions. He doesn't get bogged down in technical feasibility or "what the market says." He just tells me what he wants and why, which is literally what the best product managers in the world do.

The Hardest QA Tester I've Ever Had

Harley uses the site in ways that make total sense after he does it, but that I would never have predicted. He taps on things I didn't think were tappable. He scrolls past elements I assumed were obvious. He doesn't know to type in a filter box to narrow results. He doesn't instinctively resize his screen to see a dropdown that's rendering off the edge. He navigates the site the way an 8-year-old navigates the web, which is to say, honestly.

Every bug he finds is a UX assumption I didn't know I was making. Things that feel obvious to adults who've spent years on the internet are completely invisible to him. If a button doesn't look like a button, he doesn't click it. If a workflow requires three steps when he expected one, he gives up. If he has to scroll to find the action he wants, he assumes it doesn't exist.

We landed on a quality bar that sounds like a joke but changed every design decision I made: it needs to be easy enough that a 6-year-old could use it. Not an 8-year-old. A 6-year-old. Because if a 6-year-old can figure it out, an adult collector picking up soccer cards for the first time definitely can. That standard is unforgiving. It strips away every assumption, every "the user will know to do X," every shortcut that only works if you already understand how websites are supposed to behave. Harley enforces that standard relentlessly, and the product is dramatically better for it.

What This Actually Means

The gap between seeing a problem and solving it has collapsed. A year ago, I would have seen the soccer card pricing gap and added it to a mental list of ideas that would never get built. I don't have a development team. I don't have funding. I don't have a technical co-founder. None of that mattered.

You don't need a dev team. You need product sense, a real problem, and the tools that exist right now. If your PM instincts are sharp, AI will build what you describe. The planning, the "what and why," is the hard part. Code is the easy part now.

The question isn't whether you CAN build it. It's whether you're close enough to the problem to know what's worth building.

If you can sit next to your actual user while you build? Even better. Most PMs would kill for that kind of access. I get it every evening after homework is done, sitting at the kitchen table with a laptop and a stack of card packs.

The best product advice I got this year came from someone who can barely type on a keyboard, still needs help with his homework, is in elementary school, and plays on a U9 soccer team.

That's not a cute story. That's a lesson about where good product thinking actually comes from. It comes from caring about the problem enough to notice what's broken and being honest enough to say it out loud.