Jentic Tech Talk

Three Years of the OpenAPI Track at apidays

Three years ago, the OpenAPI Initiative (OAI) started running a dedicated OpenAPI Track at apidays events, with the idea of putting standards work in front of practitioners and keeping the community connected. Since then, we have extended the series across conference organizers. Recorded live at apidays New York 2026, this short anniversary message looks back at what that has grown into: 20 events across 10 cities and 4 continents, with 100 speakers covering OpenAPI, Arazzo, Overlays, how to leverage standards in practice, and the evolving intersection of APIs and AI. Standards only stay relevant when they show up where people are building, and apidays has been a generous host throughout.

Jentic has been supporting the effort over the past year as well, glad to lend a hand to something that benefits the wider API community. The next two stops are apidays Amsterdam (June 9–10) and apidays Munich (July 8–9), and both are still accepting a small number of late submissions. If you’d like to speak, or just come and see what the OpenAPI community is working on, get in touch.

Jentic is Heading to DevTalks Bucharest: Beyond MCP

DevTalks Romania (Bucharest, June 4, 2026) is the next stop for Jentic’s public appearances, and the talk this time is Beyond MCP: Scaling AI Enablement for API Landscapes. MCP has made it easy to connect models to APIs, and that progress is real. But in any organization with a non-trivial API landscape, exposing more tools is not the same as scaling AI.

The patterns that emerge (fragmented descriptions, missing runtime discovery, over-granular APIs, ad-hoc governance around agent execution) look familiar to anyone who has watched API platforms grow up. The argument of the talk is that AI enablement at scale is mostly an API architecture problem, not a model problem, and the path forward borrows directly from what the API community has already learned. If you’re in Bucharest, come say hello!

Arazzo Specification v1.1 Is Here

The headline addition in Arazzo 1.1 is AsyncAPI support — and it changes the scope of what Arazzo can describe. Workflows can now span both synchronous and asynchronous APIs in a single machine-readable document: call an HTTP endpoint, publish to an event bus, wait for an acknowledgment, chain the result into the next step. That’s a complete interaction pattern, fully declarative, no glue code.

Beyond AsyncAPI, 1.1 delivers: chained workflow execution with proper input mapping via Action Objects, a new Selector Object for fine-grained data extraction, better OpenAPI 3.2 alignment, identity-based cross-document referencing, and a set of clarification improvements that address real questions from practitioners running 1.0 in production.

The ecosystem has grown considerably since 1.0. Editors, validators, parsers, resolvers, generators, and standalone execution engines are all available. If you’ve been watching Arazzo from the sidelines, 1.1 is a good moment to engage.

Why AI Agents Need Deterministic API Workflows

The Nordic APIs team, Art Anthony and Bill Doerrfeld, did a solid job covering the key argument from my recent PlatformSummit talk. LLMs are powerful precisely because they figure things out. But for many enterprise use cases, you already know how things should be done. Why burn tokens leaving execution to chance? We don’t need to hard-code MCP tools and hope agents fill the gaps.

Capabilities can remain freely discoverable, through MCP, APIs, CLIs, whatever medium best suits the agent or developer, while execution stays controlled and assertable. That’s the role Arazzo plays. As an open standard from the OpenAPI Initiative, it gives you a well-defined, machine-readable mechanism to codify exactly what should happen. Agents discover capabilities freely, your business logic executes with precision, and none of it is locked into a vendor’s walled garden. For known capabilities, discoverability without determinism is just controlled chaos.

APIOps Conference Helsinki

I’ll be representing Jentic at APIOps in Helsinki on June 3rd. Three sessions and a closing keynote, all centred on what APIOps teams need to get right as agents become the dominant API consumers.

Beyond OpenAPI: Workflows That AI Agents Can Actually Execute

Arazzo 1.1 in context, and why workflow definitions are the missing layer between API capability and reliable agent execution.

You May Have OpenAPI, But Is It AI-Ready?

Hands-on scoring and improvement using agent skills combined with open-source frameworks and tooling. Concrete, actionable, no hand-waving.

Closing Keynote: APIOps in the Age of AI Agents: What to Prioritize Next

A strategic roadmap for APIOps teams navigating the shift from human developers to agents as primary API consumers.