AI on your terms: building an open and sovereign AI platform

It usually starts with one team and one API key. A year later, every team has its own keys, providers and rules, and nobody really knows what's running where or what it costs. And then the bigger question: who actually controls your AI, when models, prices and even access are decided by someone else? In this talk we'll show how one central platform fixes both: a single gateway for governance, guardrails, smart routing and failover, with each workload running where its data belongs, from hyperscaler to your own GPUs. We'll also peek at what's next, agents and sandboxes, and at the open-source standards making all of this possible.