Stop Asking AI to Do Your Old Job Faster

Early in my weekly notes to my team, I wrote an edition called "The opportunity hides in repetitive work". The argument was simple. Look for the task you do every day, the one that eats time and nobody enjoys. That is where AI pays off first.

I still believe that. But I now think it was half the story, and the less interesting half.

Efficiency and opportunity

Efficiency AI helps you do the work you already do, for less. Draft the email. Summarise the meeting. Reconcile the report. Every one of those is worth doing, and most teams are nowhere near done with them.

Opportunity AI is different. It is the work that was never on the list because it was never worth a business case. It needed a specialist, a vendor or a budget line nobody would approve for something that might not work.

Picture every idea your team has ever had as sitting somewhere on a scale of cost to attempt. There is a line on that scale. Below it, you try things. Above it, you write them off. Efficiency lowers the cost of things already below the line. Opportunity is what happens when the whole line moves, and ideas that were above it drop below.

The trap is treating efficiency as the whole strategy. Speed on the old list will be expected soon, by customers and by the people we work for. It will not set anyone apart for long.

This site is the example

You are reading the clearest example I have.

Gem Alpha would never have earned a business case. A weekly essay site with a skim brief, a steal box, audio and share cards is a small product. Not long ago it would have meant a designer, a developer and months of evenings I do not have.

Instead it went from an empty repository to live on the internet in about a week, built with AI agents doing most of the typing while I made the calls. It is not finished. Parts of it are still rough. But it exists, and a year ago it would not have.

That is the opportunity loop. Not doing an old job faster. Doing something that was simply not on the menu.

Name the port

"Go and find opportunities with AI" is useless advice. Nobody walks around with a list of inventions they have not thought of. We carry a short list shaped by our role and what our peers celebrate.

What works better is a port: take a move another craft already uses and ask what the equivalent would be in your world.

Gem Alpha is a port. The shape comes from a daily AI news show I listen to: a quick brief up top, one idea, something you can steal. I did not copy the show. I took four of its mechanics and left the rest behind: the daily pace, the news desk, the host's voice. A weekly essay about my own work does not need any of those.

The port is what made the week possible. I was not inventing a format from a blank page. I was adapting one that already worked, and asking agents to do the building. That is the pattern I would give any team: find the move that works somewhere else, strip it back to the mechanics, and try it on one real problem.

The other habit I would give any team is to stay hyper-curious with these models. Grok Bot, Meta's Muse agent, Copilot and others are now effectively running harnesses. They can go out and do good work, not just answer a question, and that is exactly what moves the line on cost to attempt. Peter Tonagh made the point that the model is only the mirror: it reflects what you bring to it. Apply that to every task, every day, in every role. Do it daily, out of curiosity, and you will get real use from the tools.

The 30-day rule

The rule I am running for the next month is simple. Every week, two loops.

One efficiency loop. Something you already do. Make it noticeably cheaper or faster with an agent.

One opportunity loop. Something you would not have tried last year. Give it four hours at most. An ugly first version counts.

If the opportunity loop fails, you have learned where the line really sits. If it works, you have moved it.

Why this matters for teams

I told my team once that AI-assisted individuals become more productive, and AI-assisted teams change how work gets done. Efficiency is the individual story. Opportunity is the team story, because it changes what the team is for.

This edition was prompted by a long-form episode of The AI Daily Brief on opportunity AI. The argument here is my own port of it.

What is one idea your team wrote off because it cost too much to try, and what would it cost now?