Species Zero

What Species Zero Taught Me

By Joey Bess··3 min read·
Lesson

I started this project on June 13, 2026, with a name I later changed.

At first it was called Genesis Beasts. Now it is Species Zero. The idea is simple enough to explain: a mobile-first game where you raise strange, one-of-a-kind AI-generated creatures and watch them evolve.

Simple to explain, yes.

Not simple to build.

I am 50 years old. I do not have a coding background. I am not sitting here pretending I suddenly became a software person because an AI tool helped me make screens and buttons. Mostly I am a curious person who kept asking, “Can this thing do one more step?”

The answer was often yes.

But not always in the way I expected.

One of the first lessons was that software can fail in ways that look completely different from what is actually happening. I had an account creation problem where it looked like the signup had failed. Very encouraging, as you can imagine. Except the account had actually been created. The app was just not making that clear.

That taught me something basic: a working thing can still feel broken to the person using it.

Then there was the missing piece with a name I still do not enjoy typing: @radix-ui/react-tabs. That caused 504 errors, which is a fancy way of saying the app just sat there and failed to load properly. I did not understand the cause at first. I just knew the thing was broken again.

This is the part I would tell someone like me: do not mistake confusion for failure.

Confusion is just part of the rent.

The biggest design lesson came from my own frustration. At one point I said, “I don't understand what is next after feeding.” That was not a bug in the technical sense. The creature could be fed. Things were happening. But I, the person playing, did not know what I was supposed to do next.

So we added a Next Step Hint. Just a little guidance to expose the game loop. Feed, wait, grow, evolve, discover. That one change taught me that games need to speak plainly. So do apps. So does almost everything.

Later, I added Stripe payments for pay-to-accelerate and a Discoverer Ranks system for prestige. Those sound like “real app” features, and I suppose they are. But adding them also made the project feel more serious, which brought a new kind of pressure.

Then I hit the wall that probably matters most.

I looked at the game and said, “This game is not good.”

That was not me being dramatic. It was me being honest.

The bones were there, but the feeling was not. It had systems, but not enough life. So the project shifted. We added ideas like Creature Voice, where the creature can have little first-person thoughts. Overnight Dreams. Evolution drift, so the creature changes in less predictable ways. Random events. Social features. Better art direction.

That moment taught me that building with AI does not remove taste. It actually demands more of it. The AI can generate a lot. It can help connect pieces. It can fix things I do not understand. But it cannot sit in my chair and feel disappointment for me.

And then, because apparently I needed more humility, there were preview loading issues. Vite re-optimization. Drift computation hangs. More terms I would not have known a year ago and still do not fully love. In plain English: the preview got stuck because the app was trying to rebuild or calculate too much at the wrong time.

What has Species Zero taught me so far?

That an idea is the easy part.

That “almost working” can be more maddening than not working at all.

That if I am confused, a future user will be more confused.

That adding features is not the same as making something good.

That AI helps most when I stay involved instead of handing over the wheel completely.

And maybe most of all, that being older is not the disadvantage I thought it was. I may be slower with the technical parts, but I know what frustration feels like. I know when something feels lifeless. I know when instructions are unclear. Those instincts matter.

If someone like me asked whether they should try building something with AI, I would say yes. But I would also say this: expect to feel foolish. Expect to get stuck. Expect to say, more than once, “I have no idea what is happening.”

Then take the next small step anyway.

Small takeaway: the tool can build fast, but I still have to learn slowly.