MC'ing at apidays New York 2026

The yearly API conference, apidays New York, is turning into a hotbed for solid discussion on what's top of mind in the API space, and as MC I had a front row seat.

I was happy to help out this year — my job was to time keep, lead Q&A, ask questions, ensure smooth transitions between speakers. (Would love to do more of this in the future when time allows. It's a whole lot easier than being the one giving the talk, hah.)

I also took notes of the trends throughout all presentations. The clear takeaway: APIs are the new execution layer for AI agents. APIs are a form of context for agentic AI, and there are emerging tactics for setting them up for success.

API providers must reframe how they handle authorization and identity, discoverability, and understandability, in order for AI to avoid hallucinations and confusion when interacting with internal and external systems.

Still, large organizations are migrating from legacy stacks, and the core principles like spec-driven development, linting specs, and internal standards all matter just the same in this era.

While the conference talks really emphasized autonomous agents becoming the new API consumer, the hallway track drifted into more on-the-ground realities:

* What will the economic impact be when token usage rises?

* Are we becoming too reliant on Anthropic?

* Why do GitHub outages keep happening?

Another recurring subject: we shouldn't be using agentic coding tools everything. Some of the proposed use cases are downright silly, whereas programmed and deterministic solutions would be far leaner and more efficient.