Why I Am Writing About More Than Agentics
I am widening the subjects I research so my agentic work has a traceable evidence base, not just more AI commentary.
Writing archive
Writing - Page 3
Older essays on agentic systems, governance, operations, incentives, and the work of making AI useful in real organisations.
229 pieces
Essays and notes
Published archive
Page 3 of 10
An Agentic Graph Is Just A Map Of The Work
A useful plain-English way to think about an agentic graph: nodes stand for the things involved, while edges show possible relationships and transitions through the work.
But for the Grace of God Go I
Deloitte, KPMG and PwC have faced public scrutiny over AI-assisted research errors. The lesson is not to laugh at them, but to build verification strong enough that the same thing does not happen to us.
Renewable Energy Is Not Material-Free
Wind and sunlight are renewable flows. Turbines and solar panels are finite manufactured infrastructure. This is what they consume, repay and leave behind.
UK Data Centre Water And Power: The Facts
A UK evidence update on data-centre water and electricity use: what is measured, what is modelled, why peak demand matters, and what good planning should disclose.
Why The UK Needs More Data Centres
Data centres already underpin ordinary British life. AI increases the need, but Britain needs a portfolio of sites matched to workloads, power, fibre, water and local opportunity.
The AI Screwed Up. Then What?
A dental AI booked the wrong specialist. The important question was not only what failed, but whether anyone captured the event, owned the system, and turned the mistake into learning.
What Has The NHS Ever Done For Us?
What agentic organisations can learn from NHS England's patient-safety systems, learning infrastructure, and continuing struggle to turn reports into safer practice.
Can You See The Sea Yet? Building The Agentic Operating System
Part four of Agentic Operating System For Your Business: moving from agentic tasks and messaging into a visible business operating system with runners, waypoints, state, evidence, and human control.
How Long Will Agentic Work Take?
Why estimates for agentic work need uncertainty classes, several definitions of done, evidence ranges, and deliberate human review pauses rather than one confident completion date.
This Morning, Codex Felt Like Her
This morning, Voice in Codex stopped feeling like a chatbot feature and started feeling like an operating interface across projects, agents and real work.
Agents Use Browsers For The Same Reason Robots Have Hands
We built the physical world around human hands and the digital world around human clicks. Browser agents are an ingenious bridge, but the future belongs to structured tools and interfaces designed for human oversight.
The Harness Is The Employee. The API Is The Contractor.
Part three of Agentic Operating System For Your Business: choosing between the deep integration of an agent harness and the flexibility of portable API workers.
Use The Harness. Keep The Escape Route.
Use the agent harness for leverage today, but keep a tested API or local-model route so convenience does not become operational dependency.
What Does AI Cost This Month? July 2026
The July 2026 UK baseline for direct AI API prices, showing quick, general, and frontier models alongside reference workload costs and the qualifications that affect a completed task.
Alex, This Is For You
A personal note to Alex about moving from ChatGPT on a phone to a simple Restaurant Control Room in Codex on a laptop.
When A Proof Of Concept Is Really Research
A thought from OpenAI Build Week in London about uncertainty, pricing, and building where no reliable pattern exists yet.
Could One Word At A Time Make Reading Easier?
A practical experiment with Focus Reader, dyslexia, attention, and comprehension.
Operational Resilience Is The First AI Task
Before you give an agent more intelligence, decide what failure you can survive. Operational resilience is the first task, not the final check.
You Are The A In AI
AI can advise, challenge, and inform. It cannot take responsibility for the judgement you choose to make.
You Cannot Vibe What You Do Not Know
AI can help us build at extraordinary speed. It cannot reliably warn us about rules, risks, and details we do not yet know exist.
Your AI Is Not An Employee Yet
Calling an AI tool an employee does not make it ready for work. A practical safety guide to onboarding agents through bounded tasks, context, supervision, tests, least privilege, and earned authority.
Making Agentic Work Visible
Why Tony is replacing message streams and human issue lists with bounded work packets that agents can triage, discuss, approve, execute, and receipt.
The New OpenAI Toolset, In Plain English
A plain-English guide to OpenAI's GPT-5.6 model family, ChatGPT, ChatGPT Work, Codex, the API, reasoning settings, tools, skills, MCP connectors, and multi-agent work.