Everybody Lies Began Small

I started Everybody Lies on 2026-06-13, which sounds very official when I write it down.

In reality, it began as a simple itch.

I like mysteries. I like the feeling of walking into a town where something is wrong, everybody knows a little more than they are saying, and you have to sort through the half-truths. But most mystery games eventually become a memory test. Once you know who did it, the magic leaks out.

What I wanted was a mystery that could surprise me again.

Not just a different coat of paint. A different town. Different people. Different secrets. A different crime. A place where the local gossip, the motives, and the lies were not all written months ago and waiting for me to uncover them in the correct order.

That is where AI started to feel less like a novelty and more like a door.

The idea was: what if the AI could generate the town? What if every citizen had something to hide? What if the case was not a fixed puzzle, but a little simulation that came alive each time?

That sounds grand, of course. Possibly too grand. Especially for someone like me, who did not come to this with a coding background and does not have a secret basement full of technical confidence.

My first version of the idea was much smaller than the thing in my head. The early scope was just the core loop: generate a town, meet the citizens, investigate, and use an investigation board to keep track of what I had learned. Local-only, meaning it lived on the device instead of needing accounts and servers and all the scary grown-up software furniture. The avatars were going to be stylized monograms, because that felt achievable.

Even that felt big.

I remember looking at the pieces and thinking, this is not one app. This is a stack of tiny problems wearing a trench coat.

A mystery needs structure. A game needs flow. A town needs names and places and relationships. Citizens need secrets. Crimes need clues. The player needs somewhere to put all the messy information. And somehow it all has to make sense on a phone screen.

That last part mattered to me. I wanted it to be mobile-first because that is how I actually play and read and poke around with things. I did not want a giant desktop dashboard pretending to be a game. I wanted something you could hold in your hand while sitting on the couch, probably when you should be doing something more responsible.

What I hoped the AI could do was not just write flavor text. I hoped it could help create the raw material of mystery: uncertainty, suspicion, personality, contradiction.

Later, the project grew teeth. Unique town images. Different voices for each citizen. Autoplay. Smoother page transitions. All the lovely little touches that make an idea feel more alive.

And with each one came a reminder that magic still has plumbing.

Town generation broke. Voices did not work the way I expected. The browser blocked autoplay. Storage filled up because I was saving too much. I learned phrases like fallback voice logic and localStorage quota, which sound like things I would have happily avoided at 30 and somehow ended up meeting at 50.

But the origin was not technical.

The origin was wanting a mystery game that did not run out after the answer was known.

I wanted to see if AI could help build a place where everybody had a reason to lie, and where I, as the player, could feel that old mystery itch again: who is hiding what, and why?

It still feels too big some days. Maybe that is part of why I keep going.

Small takeaway: start with the itch, not the feature list.