A Chief AI Officer has two jobs. Which one does yours do?

The title Chief AI Officer sounds like one job. It covers two, and they rarely sit in the same place.

The first is putting AI into what an organisation builds and sells. The second is changing how the organisation itself works with AI: who has access, what they are allowed to do with it, and whether any of it is making a difference.

The US federal mandate is a useful place to see which job the title was written for, because it is public and specific. OMB's memorandum M-25-21, issued in April 2025, gives each agency's Chief AI Officer ten duties: promoting adoption through governance, coordinating compliance, advising the agency head, representing the agency externally, keeping the AI use case inventory, running the process for high-impact AI, advising on workforce, inventorying and releasing AI code and data, guiding AI investment, and supporting the tracking of AI spend.

Read as a list, that is mostly the second job. This is my reading of ten items, not something the memorandum says about itself, and an agency is not a software company, so "what it builds and sells" only fits loosely. The closest the list gets to the first job is the process for high-impact AI, and even that governs how AI is used rather than how it is built. None of the ten is building infrastructure. That appears elsewhere in the memorandum, among the things an agency's AI strategy has to plan for.

Where I work, the two jobs sit with different people. One group builds AI into the product. My part is the other job: helping the rest of the organisation adopt AI, shaping the architecture it runs on, and representing that work outside. Adoption and workforce line up with duties on that list, and so does external representation. The architectural work lines up with nothing on it. I will come back to that, because I think it is the most interesting gap in the list.

This is one organisation, so it says nothing about what the role usually is. I am not claiming that most Chief AI Officers split the job this way, or that they should. What I can say is that the title alone does not tell you which job someone is doing, or how the role will evolve.

If your organisation has a Chief AI Officer, or someone doing the work without the title, which of the two jobs are they actually doing, and who owns the other one? And how did they get there: a new hire, an internal promotion, a role shaped around the person, or one shaped around a gap in the business?