APIs are the interfaces between teams, systems, and, increasingly, AI agents. They need to help both humans and agents:

1. Discover capabilities 
2. Make safe decisions
3. Recover when the real world refuses to behave like the schema says it should

In this track, we’ll look beyond endpoint design to the contracts, protocols, and architectural choices that make APIs trustworthy over time. 

You’ll hear practical stories and patterns for:

• Designing APIs that remain correct under retries, partial failures, slow dependencies, and competing sources of truth.
• Treating APIs as protocols, including the critical business meaning and operational expectations that schemas alone cannot capture.
• Building interfaces agents can reliably discover, understand, and use—and recognizing when human-focused developer-experience conventions break down.
• Creating secure, reusable connector layers that let agents build against real company systems without giving generated applications direct access to secrets.
• Evolving domain models and API contracts without forcing every downstream consumer through a risky migration.
   

The goal is not a perfect API design checklist. It is a better set of questions to ask when your APIs have to survive scale, organizational change, new consumers, and a future in which many callers will not be human.