A governed multi-agent system built as a Master's thesis project.
The goal is to design a system where multiple AI agents collaborate to make progressive, governed decisions, while detecting and preventing behavioral drift through structured decision trace analysis.
The project starts with a central agent and will gradually expand into a full multi-agent architecture with dedicated specialized agents.
- An AI-powered central agent that reasons and takes actions
- Structural validation of every decision before execution
- Governance guardrails with automatic escalation
- Full observability and tracing of agent behavior
- Behavioral drift detection from decision history
This project is a work in progress and will evolve step by step.