-
-
The Guide knows. You discover.
The Guide began during OpenAI Build, where the initial concept explored whether an AI-guided mystery experience could encourage people to observe more carefully, ask better questions and think critically rather than simply search for answers.
That initial prototype evolved into The Guide, an investigative thinking platform that helps people develop how they think, not just what they know.
The experience is designed around discovery. Players investigate realistic stories, examine evidence, connect clues and make thoughtful decisions to reach their own conclusions. The central design challenge was to create an engaging experience while preserving the integrity of the investigation.
As development continued beyond the original OpenAI Build project, The Guide evolved into a production product with its own Story Engine, which deterministically controls mission facts, clues, evidence, investigation state and outcomes. The player-facing Guide uses AI for conversational expression, but within a governance framework that controls its behaviour, limits its context to relevant mission state and player progress, and validates responses before they reach the player. This allows the conversation to be adaptive while keeping the underlying investigation grounded and spoiler-safe.
AI has also become an enabler beyond the player experience. The Guide incorporates AI-assisted mission production, allowing human content creators to develop structured investigations while deterministic safety controls and human accountability remain in place. An AI COO Agent provides operational oversight by analysing production health from Google Cloud Logging and customer friction from mission_feedback, then producing a structured operational report for review.
The development process demonstrated that effective AI integration is not simply about placing a model inside a product. The more important question is where AI should have flexibility and where deterministic systems and human judgment should remain in control. This led to a deliberate separation between conversational AI and the Story Engine, with governance mechanisms established around the model.
Building beyond the original prototype also introduced challenges that do not arise in a hackathon demonstration: mission architecture, production workflows, infrastructure, user experience, feedback mechanisms, commercial operations and the systems required to support the platform as it grows.
The project has therefore evolved from an experiment started during OpenAI Build into a live investigative thinking platform, with AI deliberately applied where it can enable greater capability while retaining deterministic controls and human accountability.
The Guide knows. You discover.
Log in or sign up for Devpost to join the conversation.