I have been shifting my less official, more personal, and sometimes controversial stories over here to Kin Lane, helping make sure my stories on API Evangelist do not scare the normals. This post should get at the heart of what you were looking for when it comes to the relationship of not just AI agents with API governance, but also operations, and individual APIs—because it is all connected.
My official API Evangelist stance on artificial intelligence, including AI agents, remains consistent as it is with desktop, web, mobile, device, and network applications, or just straight integrations and automation—-I avoid going down the rabbit hole on each of these applications of HTTP APIs. Sure there are nuances to each application as well as the consumers who are interested in them. But, this is my personal opinion, which is something that is codified into the API Evangelist persona so that I remain as consistent as possible across that work–I will just pull back the curtain a little bit over here within this domain.
So, whether or not you are looking to make individual APIs, as well as the API operations surrounding them, and the API governance applied across them I take the same artifact + git approach to my guidance. From vector databases to agents, OpenAPI, JSON Schema, JSON-LD, overlays, workflows, as the policies and rules governing them, will allow vector databases and agents to connect the dots. Maximize the APIs for your existing API infrastructure to inform AI. Avoid bringing in any new tools and services, and use JSON and YAML in Git to fill in the cracks, stitch together your quilt, allowing yourself to adapt and move quickly.
Not much is changing for me with this wave of AI hustle. The game is the same. Simple HTTP APIs that are well-defined using machine-readable artifacts that appropriately define the resources and capabilities being made available, but also the operations surrounding them, and the governance on top of it. This is how you can optimize the delivery of any applications of your APIs, operations, and governance. I don’t care if it is a dashboard, automation, mobile application, pipeline, or artificial intelligence. You still need the same complete and accurate artifacts to make things work.
The API Evangelist Engineering Platform Builder provides a glimpse at the road map for helping define, maintain, automate, and communicate all the layers referenced above. I am less enthusiastic about AI gettings us where we are headed, but see in as just a single application of the digital resources and capabilities we produce and depend on to run our operations. I leave it up to my readers and customers to decide what their own balance of applications is across their operations. Hope that answers your question in regards to ensuring AI agent preparedness within your enterprise.
