Mozilla.ai - Own your AI agent: running open source agents on your terms

​Owning your AI stack isn't just about ideology, it's about building systems you can actually trust and understand. This talk shows how to run your own AI agents in practice, relying on open source code and models: through concrete examples (and some inevitable failures) it will explore what distinguishes open source agents from commercial services, and present tools and techniques that make local, custom agents feasible, not just possible.

​We'll also look at Otari, Mozilla.ai's open-source LLM gateway, as an example of the infrastructure that helps make AI ownership practical by giving developers more control over models, routing, and observability while keeping a consistent interface.

​The session focuses on practical lessons learned from building and operating open source AI systems, showing how ownership, transparency, and flexibility can be achieved without sacrificing developer experience.