What I look for before I trust an Ai tool with real work
Government research found only a quarter of UK businesses adopting Ai have any practice for managing the risks. Here are the five questions I ask before a tool touches real work.
Thinking
Notes on governed Ai and building businesses that work
Written here first, shared on LinkedIn. The canonical archive lives on this page.
Articles
Building in public versus building in private: what I actually do
Buffer doubled its job applications by publishing every salary in 2013. Thirteen years on, a capable model can clone your product from a blog post. Why I now split what I share.
Why I write my own rules before I let a cheaper model do the work
Model prices have collapsed, and the temptation is to hand work straight to the cheapest one that can do it. Here is why I write the rulebook first, and what actually goes in it.
The unsexy truth about Ai in small businesses
Government research covering 3,500 UK businesses found most Ai adopters saw no change in revenue at all. The real gains turned up somewhere far less glamorous: in their hours.
What three months of silence taught me about set-and-forget automation
Automation rarely fails with a bang. It stops quietly, and nobody notices for months. What a dead content pipeline of my own taught me about governing the automations you rely on.
Governance is the accelerator, not the brake
Leaders treat Ai governance as the thing that slows them down. In practice it is what lets a business move fast and safely at the same time.
Ai adoption is a change management problem first
Most Ai rollouts fail on the people, not the technology. Why change management is the part that decides whether Ai delivers or disappears.
Why most Ai projects fail before the technology is chosen
Most Ai projects fail for reasons that have nothing to do with the model. Here is the order of work that actually makes Ai pay off.