AI Readiness Isn't a Data Problem. It's a Process Problem.

[Vontava]

If you've been in an AI readiness conversation recently, you've probably heard some version of "your data isn't ready yet." And if you've looked at what it would take to actually fix that - the modeling workshops, the SME interviews, standing up a logical layer from scratch - you've probably also thought "we don't have the people for this."

This is a conversation I have with customers every day, and this paper is my attempt to give it a better answer.

The core idea is straightforward: the context AI needs to do useful work in ITSM isn't missing. Your organization is producing it every day, in every incident, every change, and every deployment. It's just being treated as transient - it informs the work, and then disappears when the ticket closes.

What if it didn't?

This paper presents a framework for capturing logical CMDB data as a byproduct of the work that's already happening. Not instead of dedicated modeling or governance - alongside them, or as a starting point when dedicated modeling isn't available yet. Governance is built into the framework from the start, not bolted on afterward. The mechanism is outside-in capture: structured capture at the moment of work, with inline confirmation by the person doing the work, feeding directly into the logical layer that AI depends on.

The framework doesn't require clean data to start, doesn't block AI use behind governance completion, and doesn't change the work - it changes what you do with the context the work already produces.

I'm attaching the full paper along with process guides for Incident, Change, and DevOps below. Would love to hear from anyone who's tackled this, or who's working through the data readiness conversation in their environment right now.

Special thanks to @David Stefferud for support during this process!