Home

Database Systems Under Pressure

Enterprise databases are being reshaped by AI – not gradually, but structurally. This site examines what that means for the architecture, economics and safety of systems that cannot afford to be wrong.

Recent Writing

Fingerprint rendered in glowing white ridge lines on a black background, with a void at its centre where the pattern is simply absent

The Audit Trail Was Your Ground Truth. It Isn’t Anymore

The audit trail still runs. Every commit is recorded. But agentic AI has broken the two assumptions it was built on – and the incompleteness is invisible until you need it.
A disconnected mechanical governor suspended beside glowing circular data pipelines – illustrating an enterprise data loop running without its brake

The Brake Was Human. Now It’s Gone

Classic enterprise data architecture had an implicit safeguard built into it. The human in the loop provided error absorption, audit accretion and natural rate-limiting – none of which were ever specified. Agentic AI removes the human. It removes all of those protections simultaneously.
An AI agent connecting to an enterprise database, represented as a ghostly figure passing through a security gate unchallenged

Your Database Doesn’t Know What an Agent Is

Enterprise databases were built around a social contract: every action has a human author. AI agents inherit that identity model without satisfying its assumptions – and the audit log cannot say who made it happen.
AI agent committing a database transaction without human intent or authorisation – illustrating the gap between ACID correctness and intentional correctness

Transactions Assume Intent. Agents Don’t Guarantee It

ACID assumes human intent behind every commit. AI agents expose this as an architectural gap – and the enterprise transaction model has no mechanism to detect it.

See all writing →

Blog Series

Databases in the Age of AI – a series exploring what AI agents mean for enterprise systems of record, transaction integrity, database performance and architecture

Databases in the Age of AI

AI agents are collapsing the layers between intent and the system of record. This series examines what that shift means for architecture, safety and economics – without pretending the answers are simple.
Database Performance In The Cloud – series index exploring the performance and architectural realities of running enterprise databases on public cloud infrastructure

Database Performance in the Cloud

Cloud infrastructure changes the rules for database performance in ways that aren’t always visible – until something goes wrong. A practical examination of latency, throughput and cost under real workloads.
Storage for DBAs – a series building the foundation of storage performance knowledge from first principles

Storage for DBAs

Flash changed everything about storage performance, but most DBAs inherited their mental models from spinning disk. This series builds the foundation from first principles.

Database architecture. No vendor affiliations. No easy answers.