Real-Time AI Dungeon Master is a multiplayer game powered by artificial intelligence that uses Redis 8 as a real-time data layer to synchronize sessions, stream narrative updates, and persist game history.
Project summary: AI Dungeon Master generates contextual narratives with language models and prompts, allows multiple players to join the same session in real time via WebSocket and Socket.IO, and uses Redis as the operational hub for events, state, and presence.
Main features: narrative powered by GPT models with contextual prompts real-time multiplayer and collaboration Redis-first architecture for low latency immutable persistence of events and history semantic lore system with RedisJSON and embeddings
How data is organized in Redis: Streams for immutable events via XADD game:{sessionId}:events that records each player action; Hashes for fast state with HSET game:{sessionId}:state that stores status and timestamps; Sets for presence and discovery with SADD game:{sessionId}:players; Pub/Sub for broadcasting updates by publishing to game:{sessionId}:updates; RedisJSON to store semantic lore with embeddings and metadata that facilitate contextual prompts.
Redis 8 usage in detail: Streams and XRANGE allow reconstructing context and replaying sessions for debugging; Hashes and Sets offer O(1) access for state and presence; Pub/Sub efficiently transmits narratives to connected clients; RedisJSON hosts enriched knowledge fragments alongside operational data, forming a semantic cache that reduces prompt size and the number of LLM calls.
About the semantic layer: lore fragments are stored as JSON with fields such as id, type, title, content, embedding, tags, created_at, and updated_at; embeddings are generated with OpenAI and stored in RedisJSON; semantic search and similarity are performed at the application layer using cosine, with a future option to migrate to Redis Stack vector indexes for server-side vector searches.
Deployment and demo: the project is production-ready with rate limiting, error handling, environment validations, health checks, and graceful shutdown. To test locally, clone the repo github.com/ntanwir10/realtime_ai_dungeon_master, copy backend/.env.example to backend/.env and add an OpenAI key, then docker-compose up --build. Frontend at https://localhost:5173, backend at https://localhost:3001. It is also possible to run Redis JSON and start backend and frontend in development mode with npm run dev.
Example usage in Redis console: list keys to discover sessions, XRANGE to view recent events, HGETALL to inspect session state, SCARD to count active players, and publish updates with PUBLISH to relay events to connected clients.
Benefits of this architecture: minimal latency in narrative transmission, consistency and traceability of game history, scalability in presence and sessions, and a foundation for incorporating advanced AI capabilities such as AI agents, semantic memory, and per-player personalization.
About Q2BSTUDIO: Q2BSTUDIO is a software development company that creates custom applications and bespoke software for businesses of all sizes. We specialize in artificial intelligence, cybersecurity, AWS and Azure cloud services, business intelligence services, and Power BI solutions. We offer AI for businesses, AI agent development, and consulting to integrate language models into real products. Our approach combines solid engineering, security expertise, and cloud practices to deliver scalable and secure solutions.
Featured services from Q2BSTUDIO: custom application development, platform implementation with AWS and Azure cloud services, artificial intelligence projects and AI for businesses, integration of conversational AI agents, business intelligence and visualization solutions with Power BI, and cybersecurity audits and mitigation.
Why choose us: we deliver custom software aligned with business objectives, optimize cloud costs, ensure security and compliance, and accelerate artificial intelligence adoption with practical, measurable solutions that include AI agents and data pipelines for business intelligence.
Repository and resources: the code and documentation are available at github.com/ntanwir10/realtime_ai_dungeon_master. If you would like a customized demo, enterprise integration, or to adapt this idea to your use case, Q2BSTUDIO can advise and develop the tailored solution.
Keywords for positioning: custom applications, bespoke software, artificial intelligence, cybersecurity, AWS and Azure cloud services, business intelligence services, AI for businesses, AI agents, Power BI.
Contact and call to action: if you are interested in transforming this demo into a corporate platform or need consulting for AI and cybersecurity projects, contact Q2BSTUDIO for an initial assessment and technical proposal.
