Building a Game for My Daughter with AI — Part 1: What If She Could Build It Too?

She called it "a dragon but also a dog." Purple. Six legs. Completely confident about it.

I watched her finish it, cap the marker, and move on to the next page without a second thought. The way only an 8-year-old can — zero self-doubt, infinite next ideas.

And I thought: what if that drawing could actually come to life?

Not metaphorically. Literally. What if she drew something, handed it to an AI, and got back — a creature. With a name. A personality. A little world of its own.

That was the moment this project started. She doesn't know about it yet.

But here's the thing — I don't actually want to build this for her. I want to build it with her.

The Real Project

I've been working in AI for years. I've deployed models, built platforms, managed infrastructure at scale. But I've never built something for one specific person — someone I know better than anyone — with zero margin for "good enough."

If my daughter tries this and it's boring, confusing, or ugly, I'll know immediately. There's no hiding behind user research when your test user is sitting across the breakfast table.

That constraint is terrifying. And also, honestly, the most motivating thing I've worked on in years.

But the deeper motivation isn't the game itself. It's what building it together can teach her.

We live in a world where AI is already reshaping how people work, create, and communicate. By the time she's in the workforce, the question won't be "do you know how to use AI?" It will be "can you think with AI?" Can you describe what you want, iterate on it, push back when it's wrong, and guide it toward something useful?

That's a skill. And like most skills, the best time to learn it is when you're young enough that it feels like play.

So the plan: I build the prototype alone, get it to the point where something magical actually happens — a drawing transforms into a creature — and then I bring her in. Not as a user. As a co-creator. She'll tell me what she wants. She'll argue about the colors. She'll complain that the creature's eyes are wrong.

And in doing so, she'll be doing something most adults still struggle with: clearly communicating intent to an AI system and iterating until it's right.

That's vibe coding. She just won't know to call it that yet.

Why Keep It Secret (For Now)?

Because the reveal matters.

I want to watch her face the first time she draws a wobbly circle with ears and sees it transform into a living, blinking, bouncing creature with a name and a favorite food. That moment — that specific expression — is what I'm building toward.

If I tell her now, she'll have expectations. She'll ask every day if it's ready. She'll want to help immediately (which would derail the timeline but honestly improve the product).

So I'm building the foundation alone. The scaffolding, the AI pipeline, the first working prototype. Once something real exists — once a drawing can actually become a creature — that's when I hand her the keyboard.

The secret isn't the game. The secret is the plan.

The Design Thinking

Before writing a single line of code, I spent time thinking about this from her perspective. Not as a developer. As a dad who has watched her play for 8 years.

She has zero patience for instructions. If there's text to read before she can do something, she's already moved on. Everything needs to be visual, immediate, obvious.

She doesn't "fail" at art. When she draws, she's never wrong — it's always exactly what she meant to draw. The game needs to honor that. Every drawing should produce something magical, not something that looks "corrected."

Colors mean everything to her. She'll spend 10 minutes choosing the right shade of purple. The AI should notice that. A drawing dominated by warm colors should produce a confident, energetic creature. Cold blues should give you something calm and mysterious.

The naming moment is sacred. She names everything — stuffed animals, plants, random rocks she finds outside. The moment of naming her creature needs to feel like a ceremony, not a form field.

These aren't UX principles I read in a book. They're things I've observed for years without knowing I'd ever need them.

And here's what's interesting: when I eventually sit down with her to co-build this, these will be the design principles she naturally defends. She won't use that language. She'll say "that's boring" or "I want it to sparkle more." But she'll be right. And I'll be learning from her as much as she's learning from me.

The Stack I'm Considering

I won't go deep on technical decisions in this article — that's Part 2. But briefly:

The core challenge is connecting three things that don't usually talk to each other:

• A drawing canvas (web-based, kid-friendly)
• A vision AI model (to analyze what she drew)
• An image generation model (to create the creature)

Plus a personality engine that reads the drawing and decides: is this creature shy or bold? Does it love pizza or prefer moonberries?

I'm looking at Lovable.dev for the frontend, GPT-4 Vision and DALL-E 3 for the AI layer, and Supabase to store her creature gallery permanently. The architecture needs to be MCP-ready — but more on that in Part 2.

What I can say now: the moment she sits next to me and types her first prompt into this thing — even something as simple as "make it more fluffy" — she will have crossed a threshold. She will have communicated intent to a machine and watched it respond.

That's not a small thing. That's the beginning of a relationship with a technology that will be part of her entire life.

What Success Looks Like

I don't have a launch date. I don't have milestones and deliverables.

I have two metrics.

The first: her face when she sees it for the first time. If she draws something, hands the tablet to the AI, and gasps — phase one is done.

The second: the moment she tells the AI what to change. When she stops being a user and starts being a director. When she says "no, make the wings bigger" and means it, and watches it happen.

That's the real finish line.

If she shrugs at the first prototype, we iterate. If she never wants to stop building, we ship. Either way — we build it together. And somewhere in that process, a purple dragon-dog might just teach both of us something about what it means to think with AI.

This is Part 1 of an ongoing series documenting a real project — a father and his daughter building an AI-powered drawing game from scratch.

Part 2 covers the technical stack decisions: Lovable vs Replit, MCP architecture, and how I'm designing the system so an 8-year-old can eventually co-pilot it.

Follow along if you're building something for someone you love, teaching a child about AI, or just curious what happens when an enterprise AI developer stops shipping for companies and starts shipping for his kid.