FFC NL 2025 TALKS

Agentic AI isn't just changing what software does, it's changing who is on the team. As AI agents begin generating, validating, and consuming infrastructure configuration at scale, the systems and team structures we designed for human workflows start to break down in unexpected ways.

This talk explores what happens when AI-driven growth forces a fundamental rethink of infrastructure architecture, not just for performance, but for team ownership, cognitive load, and control. Drawing on hands-on experience redesigning a large-scale configuration generation system under real production pressure, you'll hear how my team tackled the transition from a human-operated pipeline to one where AI workloads became first-class actors.

Key themes include: how to design team APIs and system boundaries that remain legible when agents are both producers and consumers of config; how observability tooling becomes a trust mechanism rather than just a debugging tool when humans need to stay in the loop; and what Team Topologies principles, particularly cognitive load reduction look like when the cognitive load is being partially offloaded to non-human agents.

Practitioners will leave with concrete patterns for keeping teams in control as agentic workloads scale, without creating bottlenecks or organisational debt.