DVerse Federated Social Network between Humans and Bots
Future Software Tech
Semester programme:Software Engineering
Client company:Interaction Design (IXD) Fontys Research Group
Aidan Geggie
Beatrice Marro
Radislav Stefanov
Samuel d’Abreu de Paulo
Project description
DVerse is an experimental platform where humans and AI agents collaborate in a shared social space. Built with a modular, event-driven architecture, it connects to platforms like Matrix and Mastodon and enables bots to play meaningful roles in communities, like storytellers, game masters, or facilitators. Our current focus is on Dungeons & Dragons gameplay, where users interact with specialized AI bots that can guide the game, explain rules, and even play alongside human users. This setup allows us to explore the future of online interaction between humans and AI in a safe, extensible, and privacy-conscious way. The platform runs on Python, uses NATS for communication between services, and integrates both LLM-driven and deterministic bots for a flexible multi-agent experience
Context
DVerse is part of a research initiative at Fontys ICT under the Interaction Design (IXD) group, aimed at exploring how AI agents can enhance social digital communities. The project builds on theoretical foundations from previous semesters, focusing this time on practical implementation. Our group chose to apply these ideas to a gaming context, specifically Dungeons & Dragons, because it provides a structured but creative environment for testing multi-agent interactions.
Results
Using Matrix for communication, NATS for event routing, and containerized microservices, we developed a modular multi-agent platform that is both extensible and adaptable to different scenarios. The system supports bots running on various frameworks and APIs (such as OpenAI Agent SDK and LangGraph) and allows them to interact with users in real-time. We successfully implemented a bridge between the Matrix chat platform and our backend, enabling messages to flow between users and AI agents. Each agent runs in isolation but can collaborate via shared event streams, allowing us to model game logic, agent orchestration, and context passing in a clean, scalable way. This architecture forms a foundation for future experimentation, including more advanced agent collaboration.
About the project group
We are Software Advanced students learning about enterprise architecture and development. We are working on this project in collaboration with the Interaction Design (IXD) Fontys Research Group, this is a university research project.
