AI is changing who consumes APIs. It is no longer just web apps, mobile clients, and system integrations. It is now copilots, coding agents, workflow engines, and LLM-powered operators that can act quickly, unpredictably, and at scale.

That shift changes API design.

In this session, Florian Boymond shares a practical framework for making APIs usable and safe in AI-driven environments. We will cover what breaks when probabilistic systems become API consumers, how to design for structured actions instead of vague prompts, and which safeguards matter most once AI can trigger real side effects across business systems.

Attendees will leave with concrete patterns for:

making existing APIs more agent-ready
handling reliability, retries, and idempotency under AI-driven traffic
adding approval gates, permissions, and auditability for sensitive actions deciding when to keep existing endpoints, when to wrap them, and when MCP becomes useful

This talk is for API leaders, platform teams, and architects who want to move from AI demos to production-grade execution.