I'm Rich Inserro. Twenty-six years building data foundations that must stand up to scrutiny — now helping data leaders extend them to what AI actually requires.
Until generative AI, data governance meant metadata, data quality, master and reference data, and lineage. Those disciplines haven't gone away. AI has made them less forgiving, not less relevant.
But they were built for a world where a person read the output. A person brings context the data doesn't carry. They know that "physician" on a trial record and "physician" in the CRM are not the same relationship — and that joining them puts a clinical relationship into a commercial dataset.
A model brings nothing. It needs the meaning encoded, not assumed. And an agent doesn't just read — it acts. Which means it has to know what it's permitted to do with a record before it does it, and that has to live in the data, not in a policy document nobody reads.
Most large enterprises have built three layers of governance. The fourth is the one AI depends on, and almost nobody has it.
The model I use on every engagement. Nothing above works without what's below.
A machine can find it, understand what it means, and know what it's allowed to do with it. Semantics, context, data contracts, access control, permitted use, privacy.
The numbers hold up. Quality monitoring, master and reference data, critical data elements.
You know what you have and where it came from. Metadata, catalog, lineage.
Someone answers for it. Ownership, stewardship, decision rights.
The bottom three are your achievement. Building them took years and most of the political capital in the room. The fourth is where AI programs stall — and it's rarely on anyone's roadmap, because it didn't exist when the roadmap was written.
Interpretability is rarely finished before the rest are stable. Which order makes sense for you is one of the first things worth working out.
Same model, different depth. Most engagements start with a Map.
Where your four layers actually stand and what they need to become, with the gaps named and prioritized. A readout your executive committee can act on and your architects can build from. Weeks, not quarters.
Design and stand up what's missing. Domain model, standards, stewardship, catalog operating model, and the semantic and permitted-use layer AI depends on. I design it and drive it. Your teams and delivery partners execute.
Interim or fractional governance leadership. I own the function, run the program, carry it through the committees, and hand it to your people in a state they can sustain without me.
Twenty-six years across financial services and life sciences — enterprise data governance, analytics, risk data aggregation, and regulatory remediation. Leader of global programs in Asia, Europe, and the United States. Different regulators, same problem: data that has to be right, and provably so.
Most recently led enterprise data governance and AI readiness at a global pharmaceutical company; before that, a partner in financial services at PwC.
The hard part was never the technology. It's the people, the politics, the change management, and the patience to bring a room of skeptical executives along. I teach rather than tell, and I'd rather grow your team than make you dependent on me.
Tell me where your data is and what you're trying to build on top of it. You'll be working with me at every stage — no handoff to a delivery team, and no fifty-person program. If that's what you need, I'll say so early and point you somewhere better.