My First Big Ask
The first big prompt I sent for Everybody Lies was not a small one.
I did not ask for a button. I did not ask for a landing page. I basically asked the AI to help me build a whole little world.
The idea was an AI mystery simulation game made for phones. Every town would be created by AI. Every citizen, every secret, every crime. The player would arrive in a strange place, poke around, collect clues, and try to figure out who was lying.
A very reasonable starter project, obviously.
At the time, I was trying to be sensible about it, even if the idea itself was oversized. I asked for what I thought of as the smallest playable version: the main game loop, an investigation board, local-only storage, and simple stylized avatars using initials or monograms instead of fancy character art.
“Local-only” is one of those phrases I have now said enough times to sound like I know what I’m talking about. In plain English, I meant: don’t worry about accounts, servers, cloud saving, or strangers on the internet. Just make the thing work on one person’s device first.
That was me trying to stay out of trouble.
What came back was both thrilling and sobering.
The builder understood the broad shape of the idea better than I expected. It did not stare blankly at “AI mystery town with lying citizens,” which honestly would have been a fair response. It produced the beginnings of an app with screens, structure, and a sense of how someone might move through an investigation.
There was a town. There were people. There was the idea of secrets. There was a board where clues and suspects could live. It felt, for the first time, less like a note in my head and more like a thing I could tap on.
That part was magical.
But it was not magic in the fairy-tale sense. It was magic in the “the rabbit is out of the hat but now it is chewing through the curtains” sense.
The first version did not mean the game was done. Not even close. It meant the game had become real enough to start breaking.
And break it did.
Town generation had problems. The AI wanted information in a very particular shape, and when the generated town didn’t fit that shape, things fell over. Later, that turned into a whole lesson for me about being too strict with AI output. I don’t fully understand the technical weeds, but the ordinary-person version is this: if you ask a very imaginative assistant to fill out a form, you may need the form to be a little forgiving.
At that early stage, I also had to accept that my “simple” request was not actually simple. A mystery game is not one thing. It is a stack of things pretending to be one thing.
You need a place. You need people. You need motives. You need secrets. You need a crime. You need clues that make sense. You need lies that are interesting but not random. You need the player to feel confused in a good way, not confused because the app is broken.
That last distinction is important.
The first big prompt gave me a foundation, not a finished game. It lived up to my expectations in the sense that it proved the idea could move. It did not live up to my secret fantasy that I would press a button and receive a polished mystery experience.
I did have that fantasy, by the way. I think most of us do when we first try these tools. We pretend we are being realistic, but somewhere in the back of the brain we are hoping for a tiny software genie.
Instead, what I got was more like a very fast, very eager collaborator who needs supervision. It could build faster than I could imagine, but it could also misunderstand faster than I could catch.
The encouraging part was that the first version gave me something to react to. Before that prompt, the game was fog. After it, I could say: yes, this part works; no, that part feels wrong; this needs to be smoother; this needs to be simpler.
That is when the project really started.
Not when it was perfect. When it was imperfect in public, sitting there on the screen, asking me what I wanted next.
Small takeaway: a big prompt can open the door, but you still have to walk through and tidy the room.
Keep reading
Everybody Lies Began Small
I wanted a mystery game that could surprise me. Then I realized I was asking AI to build a whole town full of liars.
My First Real Wall
I thought the hard part would be the idea. Then the town generator broke, and I met the less charming side of building with AI.
My First Big Creature Prompt
I asked the AI to build a whole creature-evolution game. What came back was both impressive and slightly bewildering.
My First Big Ask
I asked the AI to help me start a national physician credentialing platform. It was exciting, humbling, and a little ridiculous.