Pillar 16 · Evolve
New Zealand’s Customer and Product Data Act has switched on a Consumer Data Right, and banking is designated first. A reference architecture for the consent, API and data-quality build firms should start now.
A federal moratorium on state AI laws has been proposed, voted down and revived by executive order inside eighteen months. A field note on governing under that ambiguity without stranding your compliance spend.
Portability need not mean surrendering every useful native cloud service. A decision framework for abstracting only the dependencies that matter for exit, using composition layers where they buy real optionality and native services where lock-in is acceptable.
The Data Act’s switching duty demands functional equivalence for IaaS and data portability for PaaS and SaaS. A reading of what your provider must enable, and how to design an exit you can actually exercise before egress charges vanish on 12 January 2027.
You cannot retrofit replaceability into a system after it’s been built the wrong way for three years. The seams that let you swap a vendor later have to be designed in from the first architecture decision — and almost nobody makes that decision deliberately, because it costs a little more up front for a benefit …
Every organisation building on a specific AI model right now is making a bet on that model’s continued existence, pricing, and behaviour — a bet the model providers themselves have shown, repeatedly, they can change the terms of with a deprecation notice and a migration deadline measured in weeks. I’ve written about vendor-agnostic architecture as …
Before the Model Context Protocol, connecting an AI model to N external tools and data sources meant building N separate, bespoke integrations — and connecting M different AI applications to those same N tools meant building N times M integrations, a combinatorial problem that made every new tool or every new AI application an expensive, …
A technology roadmap built on a single vendor’s protocol carries a specific, familiar risk: what happens when that vendor’s priorities shift. MCP’s move to the Linux Foundation’s Agentic AI Foundation on 9 December 2025 was designed specifically to remove that risk — and understanding exactly what changed, and what didn’t, matters for any organisation currently …
A traditional data warehouse bundles storage, compute, and query engine into one vendor’s product — buy the platform, and all three come locked together. The open lakehouse architecture breaks that bundle apart deliberately, and the decoupling is the actual point, not just an implementation detail underneath a similar-looking dashboard. I’ve written about Apache Iceberg’s role …
You cannot negotiate replaceability into a contract after the fact. It has to be built into the architecture before the vendor is chosen — deliberate seams, placed at the right boundaries, that turn a future vendor swap from a rebuild into a substitution. I’ve written about measuring vendor lock-in before signing a contract. This is …