This is my submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.
What I Built
REALITY.EXE is an open-source AI Game Master designed to turn the real world into a playable adventure.
What if, instead of spending another hour scrolling through your phone, you could open an app, receive a quest, step outside, and complete a real-world challenge?
That is the idea behind REALITY.EXE.
The core concept is simple: AI generates the game. Reality is the playground.
Players can explore quests, head into the real world, complete challenges, submit evidence, and work toward completing their adventures. The experience is designed around a simple loop:
Discover: Find a quest that sounds interesting.
Prepare: Understand the challenge and its objective.
Explore: Switch into Pocket Mode and head into the real world.
Submit: Provide evidence of the completed challenge.
Verify: Let the Game Master evaluate the evidence when a supported AI verification provider is available.
Progress: Track quest completion and player progress.
The goal is to make technology a reason to experience more of the world, not just another reason to stare at a screen.
Demo
Live website: https://reality-exe-phi.vercel.app/
Explore the interface and the real-world quest experience.
Source code: https://github.com/Inexpert-trifler/REALITY.EXE
Important deployment note: The current hosted demo does not include a hosted AI model or database. Therefore, AI-powered verification and persistent, database-dependent features are limited in the deployed version. I am keeping these limitations explicit rather than presenting simulated AI results as real verification.
Code
The project is available on GitHub:
REALITY.EXE — Source Code
The repository contains the frontend and backend implementation. The project is designed to support an AI Game Master, quest workflows, evidence submission, and verification through a configurable AI provider.
The codebase is intended to remain open for developers who want to explore the implementation, contribute features, improve verification, or experiment with different AI models.
How I Built It
I built REALITY.EXE using a web frontend and a Python backend.
Frontend
React
TypeScript
Vite
React Router
Backend
Python
FastAPI
Pydantic
SQLAlchemy in the backend architecture
AI approach
The project is designed around a configurable AI-provider architecture, with Ollama as the planned route for running open-weight models. This makes it possible to experiment with local inference rather than designing the entire experience around a single closed, hosted AI API.
I also focused on what happens when an AI provider is unavailable. An unavailable model should not magically produce a successful verification. The experience should communicate uncertainty and avoid awarding completion for evidence that has not actually been verified.
The frontend is deployed on Vercel, and the backend is configured for deployment on Render.
One of the biggest lessons was that building an AI-powered product involves more than making a model respond to a prompt. API configuration, deployment environments, persistence, verification logic, and honest failure states all matter.
The currently hosted demo is a work in progress, and the open-weight AI integration is not running as a hosted inference service.
Why Does Open Innovation Matter?
Open innovation matters because developers should have the freedom to understand, modify, and improve the technology they build with.
For REALITY.EXE, an open AI-provider architecture creates room to experiment with different open-weight models and local inference setups. Developers can explore how quests are generated, how evidence is evaluated, and how the Game Master could be improved without being permanently tied to one proprietary model provider.
It also encourages transparency. If an AI system cannot verify a challenge, it should say so instead of pretending that it succeeded.
My goal is to build a project that other developers can learn from, fork, experiment with, and help improve.
And there is a bigger reason behind the idea: technology should not only help us do more on our screens. Sometimes, it should help us put those screens away and experience the world around us.
My Agent Session
This section is optional. If you used DevRelay and have an agent session link, add it here. Otherwise, remove this section before publishing.
Prize Categories
I am submitting REALITY.EXE for the Hacktoberfest Open-Source AI Challenge — Week 1: Touch Grass.
Team Submissions
This is an individual submission.
Conclusion
REALITY.EXE started with one question: What if AI gave us more reasons to go outside?
I want to explore a future where AI does not just generate text, images, or code, but helps people turn ordinary places and everyday activities into adventures.
The project is still evolving, and there is plenty left to improve, especially around live open-weight model integration, reliable verification, and persistent progress.
But the mission is clear: make the real world the most interesting place to play.
AI generates the game. Reality is the playground.
Top comments (0)