Back to Homepage

AI-Powered "Day in the Life" Simulator (Google AI Hackathon)

As the Game Engine Lead for a Google AI Hackathon project, I spearheaded the development of an innovative "Day in the Life" simulator, inspired by the classic Oregon Trail. This project pushes the boundaries of interactive entertainment by leveraging cutting-edge generative AI to create a truly dynamic and personalised narrative experience for each player, moving beyond static content to real-time storytelling, man's oldest past-time.

Project V

Concept & Innovation

The core idea is to immerse players in a simulated "day in the life" of a character they define, within a scenario that can be user-specified or AI-generated. Unlike traditional games, every aspect of the experience – from the evolving storyline and character dialogue to the visual scenes and background music – is generated on-the-fly utilising Google's experimental and new AI APIs. This ensures a unique, unscripted journey every time, responding fluidly to player choices.

My Role & Technical Contributions

My primary responsibility was to design, architect, and implement the core "Game Engine" – the central intelligence that orchestrates the entire simulation. This involved:

  • Advanced AI Orchestration: Developing sophisticated Python backend logic to manage the game's state and dynamically call a suite of Google's experimental AI APIs.
  • Intelligent Prompt Engineering: Crafting nuanced and context-aware prompts for large language models to ensure coherent narrative progression, character consistency, and appropriate content generation across various modalities.
  • Multi-Modal Generative AI Integration: Seamlessly integrating and managing API interactions with Gemini (via Vertex AI) for storyline and dialogue, Imagen for static images, Veo for high-fidelity video cutscenes (handled asynchronously), Chirp for natural-sounding speech, and Lyria for dynamic background music adapting to the scene's emotional tone.
  • Dynamic Content Delivery: Implementing the logic to decide what type of content (text, image, video, audio) to generate next based on game state, user interaction, and computational resource considerations.
  • System Safeguards: Incorporating Google's responsible AI guidelines, including explicit prompt engineering to prevent the generation of harmful or inappropriate content.
Technologies Used

Python (Backend), Google Vertex AI (Gemini, Imagen, Veo, Chirp, Lyria APIs), Flutter (Frontend), Google Cloud Platform (Deployment).

Outcome

This project successfully demonstrated the immense potential of generative AI in creating truly interactive and endlessly replayable narrative experiences, laying the groundwork for a new paradigm in AI-powered gaming. The hackathon prototype effectively showcased how disparate AI models can be harmonised to build a cohesive, dynamic, and engaging user journey.