Blog

Latest Post

Enterprise Context Layer · Part 1 · The Overlay Pattern September 2, 2026 · Introducing enterprise semantics without touching two decades of pipelines. Worked through the single-family mortgage loan lifecycle.

The Data Governance Agent Series

A three-part series on how a high-stakes, time-boxed data rationalization project became the forcing function for autonomous, agentic data governance, and what it taught me about freeing data stewards from the perpetual quicksand of metadata curation and ever-shifting data quality rules.

Author’s Note

This series explores how AI Agents can help preserve, scale, and govern institutional knowledge at enterprise scale. It uses a business merger as its backdrop because mergers compress the underlying challenge into a visible and urgent form.

The observations throughout this series are intended to illustrate common patterns in enterprise data governance, metadata management, and knowledge capture. The focus is on the governance challenges themselves rather than any specific organization, transaction, or business outcome.

All scale and effort figures throughout this series are representative ranges intended to convey magnitude, not exact operational metrics.

Part 1 · A High-Stakes Data Rationalization, Tens of Petabytes, and the Question That Silenced the Room June 17, 2026 · A high-stakes merger surfaced a governance question that no amount of manual effort could answer, and set the stage for everything that followed. The central insight: metadata was really a knowledge problem, not an effort problem.

Part 2 · Breaking the Manual Governance Model — coming soon The economics of catchup vs. maintain. Why manual metadata curation could not scale even with more hands. The day we told data stewards to stop hand-curating, and why deterministic automation had to come before generative AI. The central claim: the operating model, not the stewards, was the constraint.

Part 3 · From Copilot to Data Governance Agent — coming soon The first Governance Copilot: what it did well, where it stopped. The crossing to a Data Governance Agent: what changes when the system decides what needs doing rather than waiting to be asked. Human accountability, autonomous governance patterns, and a compressed view of where this heads next.

The Enterprise Context Layer Series

A two-part series on how to introduce enterprise semantics without destabilizing the analytics estate you already run. Part 1 lays out the overlay pattern that binds ontology to a working integrated data model, keeps meaning versionable like code, and stays outside the operational blast radius. Part 2 walks through practical implementation end to end.

Part 1 · The Overlay Pattern September 2, 2026 · Introducing enterprise semantics without modifying two decades of pipelines. Worked through the single-family mortgage loan lifecycle, from ontology and binding catalogue through decision traces, identity spine, and federated resolution.

Part 2 · Practical Implementation of the Overlay Pattern, End to End — coming soon Everything after the design. The binding catalogue and identity spine as institutional assets: how they get authored, partitioned by source, and owned. The operating model: who signs off on what, how roles map to a typical enterprise data organization, what a RACI looks like across a full lifecycle. Sequencing: one product type, one lifecycle segment, one question.