When AI Becomes the Product Instead of Part of the Solution

Musings from AI hackathons

I’ve participated in quite a few AI hackathons recently. I love the energy, the enthusiasm, and seeing what people are building, how they’re thinking, and why they’re reaching for AI in the first place.

Then I started to notice a trend — funny at first, then a little concerning.

Every pitch, every brainstorm, every “what if we built…” started with the same premise:

👉 “We’ll use ChatGPT/Claude to…”

👉 “We’ll plug into this LLM and then…”

👉 “We’ll use a model to do [thing that AI models don’t really do well]”

No one was asking: Should this even be AI?

Was it because these were AI-focused hackathons, and people felt like they had to AI-ify their ideas? Or is it that we’re collectively blurring the line between building on top of AI and building with AI?

One treats AI like a shiny centerpiece — something you wrap your product around just to say you used it. The other treats AI like infrastructure — one of many tools to solve a real problem more effectively.

So when someone builds a ChatGPT-wrapped website with a bit of fine-tuning, yes — it is AI. But is it the same kind of AI development as building agentic systems that automate real-world workflows end-to-end? Probably not. They both have value, but they require different levels of understanding, design thinking, and technical depth. And I wonder — should we treat them the same?

It reminded me of the early mobile app gold rush. Or when people wanted to “add blockchain” to their product just because. We’ve seen this playbook before: a powerful new technology emerges, and everyone races to build for it, not with it.

As someone who’s spent the last few years scaling product teams in AI/ML, I get the excitement. I’ve felt it myself. But I’ve also seen how easy it is to lose sight of the user. The best products I’ve worked on didn’t lead with a model — they led with a user problem. AI was just the unlock.

I kept wishing more teams started with:

❓ “What problems can be solved better with AI?”

…instead of:

✨ “What can we build with AI?”

Imagine what we could build if we treated AI not as the product, but as one of many ingredients.

I still left inspired — it’s hard not to be when you’re surrounded by people playing with powerful new tools, moving fast, and turning ideas into prototypes in a day or two. But I also left with a renewed sense of clarity:

🔧 We need more builders who see AI as a tool, not a destination.

📍 More who start with real-world constraints, and use AI to meaningfully remove them.

💡 Less hype. More problem-solving.

The tech is here. The opportunity is massive. Let’s not waste it building AI wrappers. Let’s spend the time to learn how to use it — to build better, smarter solutions.