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 built for financial services and telecom.
What I write about
Browse everything →Use case metrics, value pools, BVR and the economics of AI.
9 articles → Governed & agentic AIMechanical governance, "LLM proposes, code decides", AI security and sovereignty.
17 articles → Specialist accelerators & novel solutionsBank foundation models, graph fraud detection, formally verified rules, autonomous NOC.
17 articles → Synthetic dataMeasuring fidelity, utility and privacy; augmenting fraud; generating rare faults.
5 articles → Data platforms & strategyGoverned NL2SQL, data foundations and data product portfolios.
5 articles →Featured series
All series →The Autonomous NOC
A self-improving AI reasoning layer that turns an alarm storm into one root cause in under 90 seconds, safely enough to act on a live network.
GAD-P: Governed data access
Natural-language access to enterprise data with governance enforced by code, not prompts.
Bank Foundation Models
Why and how a bank should train a foundation model on its own event data, from privacy-first data to Nova Forge.
AI Value Strategy
Closing the gap between AI ambition and measurable value, for both customers and partners.
Browse by industry
Most read & featured
Full library →Part 2: The Architecture: building GAD-P on AWS
The problem: Your data team takes weeks to answer one question.
Why banks should train their own foundation models (Part 1 of 3)
The problem: General-purpose models do not understand a bank's own event data.
Detecting fraud in real time with Temporal Graph Networks on AWS (Part 1)
The problem: Fraud rings move faster than batch models can detect them.
You can't let an LLM touch a live network, unless it can't hallucinate
The problem: Every executive asks: "you are letting it change the live network?" They are right to worry, because LLMs hallucinate structurally.
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 time is spent chasing data rather than modelling.
Your AI programme has a token bill. Does it have an ROI number to match?
The problem: Token spend is now managed like compute (FinOps Tokenomics), but most programmes cannot show the matching return.