From AI strategy to specialist solutions that ship.
Writing from the field on AI strategy and value realisation, governed and agentic AI, specialist accelerators and novel solutions, and synthetic data, with reference architectures spanning financial services, telecom, robotics and beyond.
What I write about
Browse everything →Strategy, value realisation, ROI and the economics of AI.
16 articles → Governed & agentic AIMechanical governance, "LLM proposes, code decides", agentic AI, coding agents, security, containment and sovereignty.
33 articles → Specialist accelerators & novel solutionsBank foundation models, graph fraud, formally verified rules, autonomous NOC, robotics.
27 articles → Synthetic dataMeasuring fidelity, utility and privacy; augmenting fraud; generating rare faults.
6 articles → Data platforms & strategyGoverned NL2SQL, data lineage, data foundations and data product portfolios.
8 articles →One integrated model — CAF-AI, NIST AI RMF, ISO/IEC 42001, a maturity model and ADKAR — for taking AI from ambition to measurable value. Playbook, maturity self-assessment, workshop deck and handout, all free to download.
New here? Start with these
The why: problems, business cases and how to judge value.
- Why 95% of AI projects fail, and how to put your clients in the 5%LinkedIn · 6 min
- You can't let an LLM touch a live network, unless it can't hallucinateBuilder Center
- The Fast Path: how any bank can cut loan decisions from days to minutesBuilder Center
- Why banks should train their own foundation models (Part 1 of 3)Builder Center
- Part 1: The Problem: why enterprises can't trust AI with their data (yet)Builder Center
The how: reference architectures you can build from.
- Part 2: The Architecture: building GAD-P on AWSBuilder Center
- RCA in under 90 seconds: the nine-layer autonomous NOC architectureBuilder Center
- Detecting fraud in real time with Temporal Graph Networks on AWS (Part 1)Builder Center
- How we built a formally verified fraud rules engine on AWSBuilder Center
- Training and deploying a bank foundation model on AWS (Part 3 of 3)Builder Center
Making AI auditable, secure and sovereign.
- Mechanical governance for LLM decisionsBuilder Center
- Securing agentic remediation: when the self-healing loop becomes the attack surfaceBuilder Center
- Sovereignty is not one thing: why your regulated NOC needs more than a regionBuilder Center
- How to evaluate the quality of synthetic data: fidelity, utility, and privacyAWS Blog
Adopting AI without the enterprise playbook.
Featured series
All series →Context Engineering for AI Coding Agents
AI coding agents repeat the mistakes you already corrected and lose context when they hand work to each other. Two companion articles: building a self-improving loop in Kiro from Amazon's Trace research, and how today's agents and research tackle both kinds of forgetting.
Building Data Lineage on AWS
Most enterprises can't say where the numbers in their regulatory reports come from without weeks of work. Two articles on lineage that AI builds and people confirm: the design principle, then the build on AWS.
Startup AI Readiness
Every AI readiness framework was built for enterprises. This 10-part, research-grounded one is for startups and teams under 50: assess where you stand, decide build vs buy, ship your first agent, measure it, govern it and price it.
The Autonomous NOC
47 alarms to one root cause in under 90 seconds, safely enough to act on a live network. Twelve articles on how, why, and what it costs.
Containing the Agent, Counting the Value
By 2026, enterprise AI agents stopped advising and started acting. Eight articles on containment, plus a dashboard walkthrough: when an agent does the wrong thing, can you stop it, can you prove it, and is it even worth running, ending in a value-and-risk ledger that tests every workload six ways and never blends them into one comforting score.
Bank Foundation Models
Why a bank should train a foundation model on its own event data, and how: from privacy-first data to production and Nova Forge.
Most read & featured
Full library →Part 2: The Architecture: building GAD-P on AWS
The problem: Your data team takes weeks to answer one business question. It shouldn't take more than seconds.
Why banks should train their own foundation models (Part 1 of 3)
The problem: General-purpose models don't understand a bank's own event data, and fine-tuning only takes you so far.
Detecting fraud in real time with Temporal Graph Networks on AWS (Part 1)
The problem: Fraud rings move in minutes. Batch models notice in hours.
You can't let an LLM touch a live network, unless it can't hallucinate
The problem: "You're letting it change the live network?" Every exec asks it 30 seconds into the demo, and they're right to.
The Fast Path: how any bank can cut loan decisions from days to minutes
The problem: Bank loan decisions take about 10 days, and 95% of that is chasing data, not modelling.
Your AI programme has a token bill. Does it have an ROI number to match?
The problem: Your AI programme has a token bill. Does it have an ROI number to match?
Follow on LinkedIn to see each new article and series as it is published.