Writing

Traditional cloud bills scale with traffic. AI bills scale with autonomy — and an agent stuck in a loop spends like an intern with a corporate credit card. The circuit breakers, caching, and chargeback discipline that keep agentic spend answerable to somebody.

Training built the centralised AI factory. Inference is quietly un-building it — because serving predictions to users is a latency, geography, and cost problem the industry already solved once, for content, twenty-five years ago.

Most people use "quantisation" and "compression" interchangeably. They aren't the same thing — and knowing the difference is what separates a deliberate self-hosting strategy from cargo-culting whatever ran on someone's laptop last week.

Streaming solved access and then broke it again — eight services, $1,100 a year, and a discovery experience worse than cable. The platforms that survive won't be the ones that spent the most on content. They'll be the ones that removed the friction.

The repatriation conversation has become tribal — cloud believers vs data-centre revivalists. Both camps are wrong. The right question is workload-by-workload: where do the economics, performance, and compliance of this specific system land?

An evidence-led reference on MCP for CTOs and architects — the architecture, the transports, the real security picture, the Linux Foundation governance, and where MCP fits into your build-vs-buy strategy.

APIs are already the dominant attack surface. Agentic AI changes the threat model because the caller is now an autonomous agent — so the question shifts from authenticating the user to verifying the intent, scope, and blast radius of a call.

Governance fails when it's a document instead of an operating model. The real decisions: where governance sits, what gets a human gate, why data governance is the foundation, and how to move from principles to enforcement without grinding delivery to a halt.

Why the build-vs-buy decision is never binary, how to think about total cost of ownership, and the hybrid approach that actually works in practice.

The economics of API vs self-hosted AI inference, where the crossover point sits, and why the hybrid approach wins for most production workloads.

A decision guide to agentic architecture patterns — single vs multi-agent, sequential, routing, parallelisation, orchestrator-workers, evaluator-optimiser, and reflection — with trade-offs, failure modes, and the observability you can't ship without.