AI Architecture Workshop: London Edition
A half-day working session for the people securing real AI estates
Five focused sessions, one room, and a deliberately small audience, so you'll be asked what you think rather than just listening. Everyone attending has put security controls around real AI estates, not just written about them.
AGENDA
10:00 – 10:30: Arrival
10:30 – 10:35: Introduction and meet the Noma team
10:35 – 11:10: Foundations and Shared Vocabulary
11:10 – 11:40: Agentic Control Points
11:40 – 12:00: Morning coffee break
12:00 – 12:20: Agentic Identity
12:20 – 12:50: Runtime Governance & Enforcement
12:50 – 13:50: Lunch
13:50 – 14:25: The AI Gateway Decision
14:25 – 15:00: Open Q&A and wrap-up
15:00 – 16:00: Drinks reception
WHAT YOU'LL WALK THROUGH
Foundations & Shared Vocabulary: Get every team speaking the same language about your agent estate, from agent families and the agentic stack to tool calls and AI gateway types.
Agentic Control Points: Understand where agent activity can be intercepted (harness hooks, inference gateways and MCP gateways) and where to establish control first.
Agentic Identity: Make identity consistent and traceable across a fragmented agent estate, including how IdPs, workload identity and emerging protocols fit together.
Runtime Governance & Enforcement: Combine access control and runtime protection into a single enforceable decision at every control point, illustrated by a real-world sandbox-escape case study.
The AI Gateway Decision: Work out whether you need an AI gateway now, which type fits, and why distributed gateways need a single control plane.
WHO SHOULD ATTEND
Security architects, AI platform and engineering leads, IAM owners and the infrastructure teams who run gateways and cloud networking.
Best for organisations making gateway, agent-platform or identity decisions now, across local, homegrown and SaaS agents.