Insights
Writing & analysis
Long-form thinking on the practice of enterprise AI and data transformation — governance, Responsible AI, modernization, and turning data into measurable value.
One Version of Truth: Why Your Semantic Layer Is Your AI Moat
Your enterprise is deploying multiple AI systems. Are they all looking at the same version of the truth? The semantic layer is how you compete.
Read the essay →AI Isn't a Model. It's an Engineering Capability.
Most companies treat AI as a model they bolt onto the end of a process. The ones that get real value treat it as an engineering capability embedded in the customer journey — where the model is just one box in a much larger system.
Read the essay →Financial Health Is a Data Problem
"Improving customers' financial health" has become the mission of modern consumer banking. It's a good mission — and most of the industry treats it as a marketing or product problem. It's actually a customer-data and AI problem. Here's what that really takes.
Read the essay →The Identity Resolution Problem Nobody Budgets For
Every Customer 360 program starts with a platform decision. Most of them quietly stall on a deceptively simple question the budget never accounted for: is this the same person? Here's the part of customer data that makes or breaks the whole thing.
Read the essay →Your AI Problem Is a Data Problem
The most expensive AI failures I've seen weren't model failures. They were data failures — wearing an AI costume. Here's how to tell the difference, and what 'AI-ready data' actually means.
Read the essay →Technology Is Rented. The Team Is the Asset.
I helped a bank grow from $4B to $20B in assets. The tools we started with are mostly gone now — the warehouse, the reporting platforms, the on-prem stack. What survived every change was the organization. Here's what that taught me about building a data function that outlasts the technology it runs on.
Read the essay →Putting GenAI to Work in a Regulated Enterprise — Without Losing Control of It
Every board wants an AI strategy; every risk committee wants to know what could go wrong. The teams that win stop treating that as a contradiction — and build the controls that let them deploy aggressively because they can trust the system.
Read the essay →