Writing
Four thinking pillars, five industries. Essays on transformation, enterprise AI and architecture.
Most organisations are building better RPA, not AI
Organisations are investing seriously in AI but shipping faster RPA. The gap between automating and accelerating sits not in the model — it sits in three architectural choices.
The fragmented landscape is not an excuse — it is a design problem
Fragmentation in the AI landscape is not a temporary problem the market will resolve. It is a structural condition — and how you treat it decides whether AI becomes a strategic advantage or an expensive mistake.
Pillar 1: Model Quality — choosing the right model is a strategic decision, not a technical one
Most organisations start with the wrong question: which AI model is the best? The right question is: best for what? Model quality is a strategic choice, not a technical one.
Pillar 2: Harness Strength — the orchestration layer decides whether your AI agent does work or merely simulates it
Everyone has a demo. Almost no one has a deployment. The gap between impressive AI and production-grade AI is not a matter of better prompts — it is a matter of architecture.
Pillar 3: Persistent Memory — AI without memory is a colleague who has forgotten every morning what was decided yesterday
Persistent memory is the pillar most consistently underestimated, most often deferred, and most decisive for long-term value. Without memory, an AI system learns nothing.
The three pillars together: the AI flywheel
Each pillar has value on its own. Together they create a flywheel — a system that accelerates as it does more, and improves the longer it runs.
From framework to first step: an honest approach
95% of generative AI pilots fail. Not because the technology does not work, but because organisations start with the tool instead of the diagnosis. An honest approach for breaking that pattern.
The digital transformation of commodity trading: navigating the road ahead
ERP, big data, machine learning and LLMs are reshaping commodity trading. The winners are those who combine these tools with data integrity and human judgment.
The power of daisy-chaining processes
Business is a relay race. The lead is won or lost not in the sprint but in the handoff. Digital transformation is about fixing the seams between the runners.
Your (digital) business transformation might be late — but never too late
In integrated commodity trading, end-to-end digital mastery is not a luxury. It is the difference between scaling into the boom and scrambling through it.
You're doing Agile all wrong
Agile is an adverb, not a noun. It is most useful where uncertainty is highest — and wasted where the path is already clear.
Business transformation isn't about reducing
Transformation is not about trimming the existing plant. It is about designing the outcome first, and then adding only what is needed to get there.
Why your projects fail
Project plans aren't the game. They are the manual. Too often the operation is a success, but the patient dies — because managers manage the list and not the outcome.
Stop chasing transformation
Transformation is not a goal. It is the difference between start point and endpoint. Chase the outcome, not the journey.
The excitement, the fallacy and the big steal: thoughts on ChatGPT
ChatGPT is a fantastic time saver, but anyone expecting genuine novelty from it is making a category error. A note on creation, originality and what these engines cannot do.
Shipping, Amazon and a wide blue ocean of customers
Amazon's investment in Beacon.com is not a technology play. It is a customer-centric play — and that is what should rattle the shipping industry.