We’re back in November for another evening of Azure, AI, cloud engineering and plenty of real-world lessons from people actually building this stuff. This month we’ve got two very different sessions covering some increasingly important challenges: how much AI agents really cost once you move beyond the demo, and how to build secure multi-tenant Kubernetes platforms without spinning up a cluster for every customer. **Session 1: Dieter Gobeyn - What does an agent actually cost? Designing for token economics before you ship** AI agent demos are everywhere, but there’s one question that often gets skipped: what does an agent actually cost to run? Unlike a traditional REST call, the cost of an agent can vary depending on reasoning depth, tool calls, retries and how much context it carries around. Dieter will dig into what you’re really paying for when building agents in Microsoft Foundry, explore architectural approaches for keeping spend under control, and importantly, look at when not to build an agent at all. **Session 2: Jaseetha Satkurunathan - Multi-Tenant Kubernetes: Isolating Customers on a Shared Cluster** A Kubernetes cluster for every customer gives you isolation, but it can also mean more infrastructure, more cost and more operational overhead. So what happens when you want multiple customers to safely share the same underlying Kubernetes platform? Jaseetha will explore a practical approach to designing a multi-tenant Kubernetes environment, looking at how customer workloads can remain isolated while avoiding unnecessary duplication of infrastructure. This session is aimed at cloud, DevOps, platform and Kubernetes engineers thinking about multi-tenancy or looking for more efficient ways to host customer workloads. **Meetup location** what3words:///daisy.coffee.awards **Agenda:** 6:00 - 6:30 - Welcome & networking 6:30 - 7:15 - Foundry IQ and the Journey to Agentic Search 7:15 - 7:45 - Food, drinks and general chatter 7:45 - 8:30 - Stop building data lakes for

Description supplied by the event organiser.