M-A-R-I: your AI agents, working as one

 

M-A-R-I: your AI agents, working as one

An open-source framework to build, orchestrate and scale multi-agent AI systems across a distributed cluster. AI is no longer a single model answering a single question. Real value comes from many specialised agents, each doing one thing well, that collaborate, react to events and share what they learn. Building that by hand means fighting with distribution, scheduling, state, failures and scaling, long before you write any business logic. M-A-R-I takes care of that foundation. Built on .NET 8 and Microsoft Orleans, it gives you a proven distributed runtime and a simple, fluent way to describe how your agents work together. You focus on what your agents do; M-A-R-I makes them run, together, at scale.

https://github.com/kaonomi-net/mari