Essays, playbooks, and measured findings on building with AI, running agencies, and how work actually gets done - plus personal stories and the opinions I will defend. Much of it published here first, and all of it open.
For twelve years my agency ran the same check on every brief before we quoted. It caught the projects that would have hurt us. It never had a name, so nobody outside the building could ask for it, repeat it, or pay for it on its own. The check was real. Without a name it was invisible.
September closed on an open question: whatever moves a page from cited to named, the run had not measured it. October is the first data point on that question. It is also the first run that failed partway, and the failure was mine. Both belong in this report.
Every agency owner I talk to about AI gets to the same question within ten minutes: which tools should we buy? It is the wrong question, and one sentence shows why. Your competitor can buy the same tools. The thing they cannot buy has been piling up on your server for ten years.
My previous co-founder and I started our digital agency in 2009 because we worked well together. Nobody told me that running an agency is a sequence of different jobs wearing the same title, rotating every few years, each one asking a different version of you to show up. The partnership was built on the shape of the first job. The job kept changing. We did not.
A few years ago, if you had told me I would be co-building a coaching program for agencies, I would have been polite about it and changed the subject. What changed my mind was noticing the gap between the methodology coaches sell and the methodology they actually use, and meeting a partner whose background is the exact opposite of mine. Software refuses to accept fuzzy. That refusal turns out to be the whole point.
The normal way to build a coaching business is methodology first, platform later, as delivery infrastructure. We are doing the opposite: the methodology and the software are emerging together, and each one refuses to accept the other's vagueness. Eight months in, the work has taught us what the methodology actually is, and why an operating environment is a different category of product than a course.
Ask what the most valuable system in an agency is and you will hear about the CRM, the project tool, the pipeline. Ask what actually runs the place and the honest answer is a skull. The standards, the client history, the judgment, the memory of what was decided: all of it commutes to the office inside one person. Twenty years of fixes failed to move it. I think I finally understand why, and what moves it instead.
Agency AI pilots die in a drawer because the task was chosen by excitement instead of by fit. The right first automation is almost never the most impressive one. It is the most repeatable one, and repeatability can be scored. This is the triage I run before any agency automation project: where the candidates hide, six criteria to score them, and the pilot rules that keep the winner alive. Automate the work around the craft, never the craft.
Last month closed with a deferral: the site had been restructured between runs, no effect was shown or expected, and indexing lag meant any effect belonged to September or October. This is that sentence being answered, at the early end of its own window.
In 2015 I was running the exact version of the agency everyone now predicts the death of. Thirty people, a pyramid, margin living in the gap between what a junior cost and what a junior billed. Predictions about the industry are usually written so they can never be wrong. These five are written the other way: dated, specific, and made to be graded in 2029. Hold me to them.
I write regularly on LinkedIn about what I'm building and learning: agency growth, AI development, product judgment, and the messy reality behind making things work.
Follow on LinkedIn