Blog

Long-form essays on data architecture, lakehouses, governance, and AI.

• You Can't Buy Neutrality — Why the clearinghouse for AI agents will be chosen, not imposed — and why the incumbents can't hold the seat. Captivity needs no structure; trust requires it.
• Separation of Powers Is a Security Pattern — What 250 years of constitutional democracy can teach us about building trustworthy data systems: three separated powers, a constitution above them, and checks that make it structurally hard for any component to go rogue.
• The Back Door Just Got a Deadline — For three issues this newsletter described an architectural gap. Now BCBS 239, DORA, and the EU AI Act have turned governing what AI agents read from unstructured documents into a regulatory deadline for banks.
• The Evolution of the Warehouse: From Gatekeepers to Trust Infrastructure — How warehouse management evolved from the Catalog to the Metastore to the Control Plane — and why Trust Infrastructure for the ungoverned 80% is the next inevitable primitive.
• The Two Universes Inside Your Data Lake — Data lakes were built for facts. AI runs on meaning. The deterministic layer we spent a decade governing and the ungoverned probabilistic layer where decisions actually get made are two universes — and nothing connects them.
• Did We Forget the Lake? — Structured tables get all the attention while 75% of data is unstructured. A case for a volume-centric lakehouse control plane to govern the unstructured Nebula.
• The 2025 State of the Apache Iceberg Ecosystem Survey Is Out — And There's a Lot to Unpack — A reaction to the 2025 State of the Apache Iceberg Ecosystem survey: Spark dominates, the catalog market is fragmented, and the next fight is the governance control plane.
• Apache Iceberg: From File Format to Business-Oriented API — Apache Iceberg's new File Format API, shipping in 1.11.0, lets new storage formats plug in cleanly across engines, deepening Iceberg's promise of vendor-independent data.
• Open Data Immunity: Rethinking Governance for the Age of AI Agents — Open file formats and open compute are not enough. Governance must be portable, auditable, and enforced at agent speed. Introduces Open Data Immunity for the AI-agent lakehouse.
• A Simple Question: Is Your Data a Competitive Advantage for Your Company? — Most companies say data is a competitive advantage but behave otherwise. This essay explains when to buy versus build, and why open formats and enforceable governance make data a real moat.
• The Cloud Economic Cliff: Why Data Gravity Breaks SaaS at Scale — Every company hits a data economics threshold where cloud SaaS stops making sense. This essay maps the hidden multipliers and argues for owning your composable data plane.
• The Unexpected Gift: How Iceberg Benefits Both Customers and Vendors — How Apache Iceberg unbundles storage, query engines, and the control plane, freeing customers from lock-in and letting vendors specialize instead of building do-it-all stacks.
• Your Data Strategy Is Only As Good As the Business It Supports — A business-leader's view of the Lakehouse: how it turns data into a revenue-generating asset through better ROI, lower TCO, reduced risk, and faster innovation.
• The Database is Dead: The Plot Twist in Data's New Era — Monolithic OLAP databases broke under modern scale. Apache Iceberg made data the open foundation, turning databases into specialized query engines in a new open ecosystem.
• The Data-AI Control Plane: Iceberg REST Catalog's Journey to Intelligence — How the Iceberg REST Catalog evolved from a simple metadata store into an intelligent control plane that governs both data and AI assets across the modern Lakehouse.
• Don't Reinvent the Monolith: Mapping an Open Lakehouse Future — Using a Wardley Map to show why bundling catalog, compute, and storage into one platform traps innovation, and why an agnostic, open Lakehouse wins.
• Data Trapped? The Shift to Agnostic, Business-First Architecture — Why rigid, locked-in data stacks are giving way to modular, agnostic architectures built on Iceberg, semantic layers, and MCP that adapt to the business.
• Welcome to Agnostic Data Architecture — Rethinking Data Without Lock-In — The inaugural edition of Vakamo's Agnostic Data Architecture newsletter, introducing its focus on open Lakehouse patterns, Lakekeeper, metadata, and governance.