Craig Newsome
Sydney · 33.87° S — 151.21° E

Systems that hold.Stories that travel.

Enterprise architecture and AI governance by day. A travel press, a review methodology and two fantasy novels by night.

Technology & enterprise architecture

Systems that hold

Enterprise architecture, technology strategy and AI governance — two decades of making complex estates simpler, safer and cheaper. Now advising independently.

Consultancy · review governance

Tenth Chair

Code review for the AI era: nine seats, formal governance, and prompts that make AI review draw blood instead of noise.

The book

Governing Agentic AI

An architecture playbook for systems that build themselves — written from a year of running autonomous agents against real systems and real money.

Speaking & education

Say it out loud

Open to presenting at events and running education programmes — AI governance, enterprise architecture, and what agentic systems actually do when nobody is watching.

The workshop · built with AI, governed hard

Built, not just advised

Three AI-built systems running in production: Pendragon, an autonomous market-research and trading engine under pre-registered kill criteria; Atlas, an honest niche-tester for creative sellers; and Samson, the strength log that replaced two subscriptions.

Longitude Press · travel colouring books

Drawn from the road

Fifteen destinations, every plate hand-drawn from a real photograph taken on the road. No stock images. No AI.

Fiction · writing as C J Newsome

Three novels on the shelf

Echo's Call, Gaia and Britannia — rebel witches, drowned cities, and a country up for sale. On Amazon now.

2003Enterprise architecture, financial services
2019Head of Technology – Architecture, AMP
2025First autonomous agents in production
2026Longitude Press · Tenth Chair · the book

The same discipline, two very different pages.

I've spent two decades designing systems where being wrong is expensive — banking platforms, trading infrastructure, the plumbing that has to hold. The last year I've applied that discipline to a newer problem: AI agents that write code, run experiments, and report their own success. It turns out they need what junior teams and complex programmes always needed — honest evidence, adversarial review, and someone willing to say stop.

The other half of the time I'm travelling, with a camera and a notebook. The photographs become line art and the line art becomes colouring books; the notebook becomes fiction. Machines get the governance. The stories and the drawings stay human.