← All Articles

The Best of Insights 2026

17 August 20267 min readAI StrategySharePDF

Executive Summary

89 articles since January. Most were written to answer one specific question a client or reader had that week, then move on to the next one. A handful turned out to matter more than the week they were published in. This is that handful, organised by the question each one answers, not by when it went up.

Core conclusions

  • Twelve articles, grouped into six recurring questions: is the organisation ready, which tasks are safe to delegate, why the demo breaks in production, what a working agent programme looks like, what’s actually changing in how people work, and where the risk is hiding.
  • None of these are the newest posts. Recency and usefulness rarely line up, and this list is deliberately biased toward the second one.
  • New here? The shorter, ordered version is Start Here, nine of these as a single reading path instead of six grouped questions.

I write most weeks, and most of it is timely by design: a reaction to a client conversation, a new regulation, a number that came out that quarter. That’s the right way to run a blog, but it means the archive rewards browsing less than it should. This is the fix: twelve pieces that answer questions I still get asked today, regardless of when they were written.

Is your organisation actually ready?

Before anything else gets evaluated, procured, or piloted.

Which tasks are safe to hand to an agent?

Not every task belongs to an autonomous system, and the ones that don’t will tell you before you deploy them, if you ask the right questions first.

Why does the demo break in production?

The gap between an impressive prototype and a system that survives contact with real data and real users.

What does a working agent programme look like?

Past the pilot, into something a Chief AI Officer can actually run and defend.

What’s actually changing in how people work?

Less about job titles disappearing, more about which tasks inside a job are worth doing yourself anymore.

Where is the risk hiding?

The two governance gaps I get asked about most, precisely because most policies were written before either one existed.

If this is your first visit, the Start Here page walks through nine articles as a single ordered path instead of six grouped questions. For a live application of the frameworks referenced above, the AI Governance & ROI Executive Programme runs half a day, not eight months of trial and error.

Was this useful?

Read next
AI Could Do the Job Tomorrow. The Math Still Caps How Much You Can Automate.
AI Strategy

AI Could Do the Job Tomorrow. The Math Still Caps How Much You Can Automate.

A corrected scaling law, real data on correlated model error, and the economics of compute scarcity all point the same way: near-total AI capability does not translate into near-total substitution. The five numbers that actually set the ceiling, and the one most enterprise business cases quietly leave out.

AI Companies Actually Printing Revenue in 2026
AI Strategy

AI Companies Actually Printing Revenue in 2026

Not cloud infrastructure, not copilots bolted onto existing software. A benchmark of the AI-native companies whose entire business is the AI product itself, what they charge for, and why the pricing model is the real signal to watch.

Parenting a Superintelligent Child: What Values Are We Actually Passing Down?
AI Strategy

Parenting a Superintelligent Child: What Values Are We Actually Passing Down?

Elon Musk told SpaceX employees they'd 'effectively be the parents' of Grok. Take that seriously and sixty years of parenting research becomes strangely relevant to AI governance. What decades of studying real children tell us about instilling values in something that will eventually outgrow every rule we write.

Terence Kok
Before You Go

I went back and forth on whether a 'best of' post is a bit self-indulgent, then remembered how often I get asked for a reading list by someone new to a client conversation. This is that list, finally written down instead of retyped from memory every time. If your favourite didn't make the cut, that's a fair complaint, and also exactly what the contact form is for.

Terence Kok