Prologue
August 9, 2026
Today was one of those days where a vague idea started turning into something that feels like a real game.
I’ve been exploring an AI-driven interactive story engine called The Fifth Voice. The initial concept was essentially an AI-powered choose-your-own-adventure game, but working through the mechanics today revealed something much more interesting.
The core idea is now this: four characters sit around the table and tell you what they think you should do. You make the final decision.
Each character has their own personality, skills, fears, beliefs, relationships, and evolving perspective. They won’t necessarily agree with each other—and they won’t necessarily agree with you.
And there’s always a fifth option: Try Something Else.
The player can type what they want to attempt in their own words. AI interprets the idea, but a deterministic game engine decides whether it’s actually possible and what consequences follow.
That distinction feels important:
The AI can propose. The engine decides. The AI tells the story.
It gives the player a lot of freedom without turning the game into an anything-goes AI narrative.
The flagship example is a Cretaceous survival game involving four high-school students who become stranded millions of years in the past. The concept evolved today from simply trying to escape to something I find much more compelling: what if they eventually realize they’re not going home?
The story becomes:
Survival → Exploration → Adaptation → Settlement → Sustainability
They’re not just trying to stay alive. They’re trying to build a future.
That also changed how I’m thinking about endings. Instead of simply surviving for X number of days, the game can end when the group reaches a meaningful outcome—perhaps they’ve established a sustainable settlement, perhaps the group has fractured, or perhaps they’ve survived but failed to build a viable future.
The characters themselves should change along the way. The cautious kid might become a leader. The tech kid might become a builder. Relationships can strengthen or fracture. The Council at the end of the game should feel different from the Council at the beginning because the people sitting around it have changed.
Technically, the architecture is becoming much clearer too: a deterministic TypeScript game engine underneath, with AI operating around it as an interpreter, character voice, and storyteller.
But the biggest decision today was actually about scope.
I’m not going to try to build the entire no-code platform first.
The plan is:
The first goal is a small 15–30 minute vertical slice that answers one question: is this actually fun?
And the sentence that best captures what I’m trying to create is probably:
“I don’t know what will happen next. But I know it will happen because of me.”
That’s the part I’m most excited about.