I have never been short of ideas. Ask anyone who has worked with me, studied alongside me, or had the misfortune of sitting next to me when something occurs to me at 11pm. Ideas were never the problem. What has held me back, my whole career, is the gap between the idea and the thing: a lack of skill, a lack of time, and usually both.
So I have spent weeks on little sleep researching the exact thing I wanted to build. Crafting the campaign. Hand-coding lead-generation hooks in JavaScript. Wiring everything into the CRM so the follow-up and the reporting were clean. And still, often, a compromise. A deadline that forced a shortcut. The quiet dissatisfaction of watching something land at eighty per cent of what I had seen in my head.
When AI arrived, I assumed it was a fad. I had dabbled with Gemini, Copilot and ChatGPT, and the results were always the same: enthusiastic, confident, and generic AI slop. Handy for shortening a paragraph into a tight line for an email. Useless for anything that needed a point of view. I filed it under hype and moved on.
Then, on a cold March day, a trusted friend and colleague, Emma Aiello, who runs the B2B marketing consultancy Red Unicorn, told me to try Claude. I was polite about it. I did not expect much.
What changed my mind was an article an ex-teammate sent me. Reading it, the penny finally dropped, the thing everyone had apparently been going on about. The part of marketing I had always found hardest was never the thinking. It was the doing. I was never short of angles or content ideas; I was short of the hours and the craft to research and write them all to the standard I wanted. And that was precisely the part that had just become possible.
Here is the moment it became real. I had just started a new role, the second marketing hire, and a crisis was unfolding in a part of the world that needed a response from us. My boss asked me to draft something and send it over for review. I felt the familiar cold panic of a blank page against a clock. So I turned to Claude. I gave it a proper brief: the audience, the channel, the sentiment, what we needed to say and what we could not. The draft came back good. I tweaked it, sent it on, and half-expected the question. Would he ask if AI had written it? Was this cheating? The post was signed off and live within twenty minutes. Something I could have procrastinated over for hours was done in minutes, and it was not slop, because the brief was the guarantee.
The bigger unlock was mine, not the tool’s. I had spent years developing my own marketing methodology. Now I could build it out into a proper AI marketing system, founded on my own thinking rather than someone else’s template. The methodology was what made the AI useful, not the other way round.
I now run three AI marketing systems for three different businesses, each built on that original methodology. The production layer has shrunk dramatically. What used to take weeks can take days, sometimes hours.
Here is the honest part, though. It has not stopped the late nights. The ceiling that used to sit on my skills has broadly gone, which leaves only two limits: time, and my own standards. Those two were always the point. I still deliver campaigns into the small hours, because I still care whether they land at a hundred per cent rather than eighty. That has not changed, and I would not want it to. It is the reason I love this job.
I held out longer than most. I called it a fad, then I called it slop. I was wrong twice. And the thing that turned the slop into something worth drinking was not the machine. It was having a methodology of my own to pour into it. The Kool-Aid only works if you bring your own recipe.
