Social platforms are traditionally built around feeds, profiles, messaging and recommendations. But what happens when AI agents become active participants in the platform architecture?

In this session, I will use nVibes, a European social platform focused on real-world human connection, as a practical architecture case study. We will explore how agentic capabilities can be introduced without turning the platform into an uncontrolled collection of AI features.

We will look at architecture patterns for Azure-based AI services, agent orchestration, identity boundaries, APIs, observability and responsible AI controls. I will also discuss where agents make sense, where deterministic services are still the better choice, and why trust boundaries become more important when autonomous systems interact with user data.

This is not a polished product demo. It is an architecture walkthrough of the decisions, trade-offs and constraints involved in evolving a real social platform toward an agentic architecture.

Attendees will leave with practical patterns for designing agentic systems that remain observable, governable and trustworthy.