Writing on agentic AI
Short, practical pieces for people who intend to build something rather than form a view.
All guides
Start here
What the ideas are, and what the words actually mean.
What can you actually build in one day with AI?
More than a demo, less than a product. What a single day of hands-on AI building realistically produces, and the four things that will not fit in it.
Read August 2026A second brain for AI is not the note-taking kind
The term comes from personal note-taking, but a second brain your AI can use is built differently. What changes, and whether you need both.
Read October 2026Scout, drafter, reviewer: the workflow pattern that does real work
The three-step agentic workflow pattern behind most useful business automation - what each step does, why the split matters, and where the human sits.
Read October 2026A plain-English glossary of agentic AI
The terms that come up when people talk about agentic AI - agents, workflows, orchestration, MCP and the rest - defined without jargon.
Read July 2026Agentic AI for business leaders: where to start
A practical starting point for senior leaders who already use AI and want to build agentic workflows that do real work.
Read October 2026What is an agentic AI bootcamp?
What actually happens at an agentic AI bootcamp, who it is for, and how it differs from an AI course.
ReadChoosing how to learn
Formats, claims, and what to ask before booking anything.
Getting a one-day AI bootcamp approved, and what to promise
The request that gets a day out signed off names one task, one deliverable and one date. What to put in it, what comes back, and what not to promise.
Read October 2026Who should take the seat - you, or someone in your team?
One seat, two candidates. Why this is a question about who owns the task rather than who is most senior, and the three tests that settle it.
Read October 2026Should you run an in-house AI day, or send people to a cohort?
Most providers only sell the in-house version, so the comparison is rarely written down. What each format is good at, and the question that settles it.
Read October 2026How long does it take to build a second brain for AI?
The usual answer is weeks of capture habit. A second brain for AI is built differently - a working one in a day, then minutes at a time to keep it current.
Read September 2026Claude, Cowork, Claude Code or Codex: which one for which job
Four AI tools, overlapping names, and most comparisons are written for developers. How to choose if your work ends in documents rather than software.
Read October 2026Three different things get sold as a ‘second brain’
The phrase is attached to at least three unrelated products, all bookable, all using the same two words. How to tell which one you are looking at.
Read October 2026When do AI executive cohorts run, and when should you go?
Executive AI cohorts no longer cluster by season - most run online several times a year. What that changes about when to book, and what to do between them.
Read October 2026Nine questions to ask before booking AI training
A buyer’s checklist for AI courses and bootcamps - nine questions worth asking any provider, and what a straight answer to each one sounds like.
Read October 2026You’ve finished an AI executive programme. Now what?
The gap most AI executive education leaves - and how to turn a strategic understanding of AI into something running in your own work.
Read October 2026AI bootcamp, AI course or executive education: which do you need?
The three main ways senior people learn AI, what each is actually for, and how to choose based on what you want to walk away with.
Read October 2026What does ‘hands-on’ actually mean in AI training?
Almost every AI course calls itself hands-on. A practical test for what the word actually covers, from watching demos to building against your own work.
Read September 2026Why in-person beats online for learning AI
Online courses teach you about AI. A day in the room is what gets an agentic workflow built. Here is the difference, and why it matters.
ReadPutting it to work
Where it helps in a particular kind of job, and how to make it stick.
Is your AI workflow saving time, or just feeling faster?
On simple tasks, AI has been measured to save effort rather than time, not both. What that means for a multi-step workflow, and what to measure instead.
Read September 2026The agent wrote the code. Why is nothing live?
Coding agents write working code and stop. The distance between that and something other people can open is six specific things, none of them coding.
Read September 2026Getting an agentic workflow past IT, legal and your manager
The questions IT, legal and your manager will ask before an agentic workflow goes live, and how to have the answers written down before the meeting.
Read October 2026Is it safe to put company documents in an AI second brain?
A second brain is not one pasted document - it is material your AI tools read again and again. Three questions that decide what can safely go in.
Read September 2026Where the human actually goes in an agentic workflow
Most agentic workflows put the human in the wrong place, relaying text between tools. Where oversight belongs, where it is theatre, and when to move it.
Read September 2026Why your second brain stopped getting used
Capture was never the problem. Most second brains die because the reading half was left to you. What changes when the reader is a model instead.
Read August 2026A stale second brain is worse than none at all
Note systems rot into clutter you skim past. A second brain your AI reads rots into confident wrong answers. What decays, and the routine that catches it.
Read August 2026Agentic workflows fail quietly. Here is how you would notice
Software fails loudly. An agentic workflow fails by handing you a confident answer that is wrong. Four checks that catch it before anyone else does.
Read October 2026What actually goes in a second brain for AI?
The five kinds of context worth storing so your AI tools stop giving generic answers - and the things that should never go in.
Read October 2026How to choose the task you bring to an AI bootcamp
The task you pick decides what you get out of a build day. Five tests for choosing well, and the work worth leaving alone permanently.
Read October 2026How to run an AI pilot that survives contact with your team
AI pilots rarely fail on the technology. They fail because nobody owned the task, success was never defined, and week two was never planned for.
Read August 2026Agentic AI for public-sector and NHS leaders: start where governance says yes
How to apply agentic AI inside a public-sector organisation - which work suits it, which never will, and why governance is the design brief, not the blocker.
Read July 2026Agentic AI for consultants: sell the judgement, not the assembly
Which consulting work suits an agentic workflow, which does not, and how a workflow built once becomes margin on every engagement after it.
Read October 2026Agentic AI for finance teams: where it actually helps
The finance work that suits an agentic workflow, the work that does not, and where human judgement has to stay.
Read