I'm Rich Inserro. I've spent over two decades building data foundations that met business objectives and held up to regulatory scrutiny. Now I help data leaders get those foundations ready for AI.
Before generative AI came along, data governance primarily meant metadata, data quality, master and reference data, and lineage. All of that still matters today; however, AI is less forgiving of deficiencies in how data is managed.
Data governance disciplines were designed for a world where a person reads the output and could apply context the data itself doesn't carry. For example, an analyst would understand that "physician" on a clinical trial record and "physician" in the CRM describe different relationships, and that joining the two may put clinical trial information into a commercial dataset.
An AI model has no way of knowing that relationship unless someone spells it out in a form the model can use. Agents raise the stakes further because they take action. Before an agent does anything with a record, it needs to know whether it's allowed to take the action, and that rule must live with the data itself.
Most large companies have put years of work into three layers of governance: ownership, understanding, and trust. But there is a new fourth layer that AI depends on, and very few have built it.
The Unbuilt Layer is what AI depends on. It's how a model or an agent knows what data means and when it can take action.
In the future, when we look back at the winners and the also-rans in the AI race, the winners will have invested resources in building the connectivity that makes data Interpretable by machines. But this Interpretability layer needs to be built upon a solid foundation where data is Owned, Described, and Trusted.
If you've built the bottom three layers, that's a real accomplishment. It probably took years and a good share of your political capital. The fourth layer is usually where AI programs get stuck, and it's often missing from a data roadmap because nobody was thinking about it when the roadmap was written.
In practice, interpretability can't be finished until the other layers are stable, so deciding what to tackle first is one of the first things we'll work out together.
A machine can find the data, understand what it means, and tell what it's allowed to do with it. This covers semantics, context, data contracts, access control, permitted use, and 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 is accountable for it. Ownership, stewardship, decision rights.
Each one uses the same AI-ready model; the difference is how deep we go. Most clients start with a Map.
An assessment of where your four layers stand today and what they need to become, with the gaps named and ranked. You get a readout your executive committee can act on and your architects can build from, usually within a few weeks.
Designing and standing up what's missing: the domain model, standards, stewardship, catalog operating model, and the semantic and permitted-use layer that AI relies on. I lead the design and drive the work, and your teams and delivery partners handle execution.
Interim or fractional leadership of your governance function. I run the program, take it through your committees, and hand it over to your own people once they can keep it going without me.
I've spent twenty-six years in financial services and life sciences, working on enterprise data governance, analytics, risk data aggregation, and regulatory remediation, and I've led global programs in Asia, Europe, and the United States. The regulators were different everywhere, but the problem was always the same: the data had to be right, and we had to be able to prove it.
The hard part has never been the technology. It's the people, the politics, the change management, and the patience it takes to bring a room of skeptical executives along. I'd rather teach than tell, and I'd rather grow your team than leave you dependent on me.
Most recently I led enterprise data governance and AI readiness at a global pharmaceutical company. Before that, I was a partner in financial services at PwC.
Tell me the business problem your data is holding back, and we'll work out where to start.