There's a kind of work that happens before any work is visible. Before the first asset renders, before the first headline lands, before the output exists at all: someone spent weeks teaching a machine what your brand actually sounds like. Not what it says. What it sounds like when it's being honest. That part doesn't show up on the timeline. It barely shows up on the invoice. It's the 80% that keeps the remaining 20% from looking like it was written by a mildly panicked chatb

AI isn't dumb. It's democratically average. It was trained on billions of human fragments: LinkedIn thought leadership posts, SaaS blog articles, vendor emails asking to "just hop on a quick call", Yelp reviews for a pizza place in Omaha, and every "per my last email" ever written. It computed the gravitational center of all of it. When you ask for content, that's exactly what you get: the gravitational center of everything. Think of it as hiring an intern who read 400,000 ma

Everyone is asking for " AI fluency ". Almost nobody knows what they're actually asking for. Which is fine, honestly. The marketing industry has a long and proud tradition of demanding things it can't define. Remember when everyone needed "synergy"? Or "authentic storytelling"? At least with AI fluency we've upgraded to a term that sounds like it could mean something. Here's what I've noticed over the past year: every job post, every CMO conversation, every agency brief menti

I use it to think better, faster, and with more responsibility than I ever could alone. Over the past few weeks I've been posting more here. And yes, I've been using AI to do it. Some people liked it. Others criticized me for it. To both: thank you . Because human criticism isn't AI's enemy . It's the mechanism that keeps AI inside some reasonable limit of sanity. AI that is never challenged doesn't grow. It just repeats itself faster. But something needs to be said clearly

AI is a machine of averages . It was trained on everything humans have ever made, and when you give it a thin brief, it produces the statistically expected answer . Fluent. Competent. Indistinguishable from what every competitor with the same tools is generating right now. That's not a bug. That's the architecture. The danger isn't bad output. It's plausible output. Content that passes every internal review, checks every box, and slowly drains a brand of the one thing that ac

For the last 3 years, I’ve built and orchestrated entire AI ecosystems. I’ve spent most of my days managing more autonomous agents than actual people. And here is the uncomfortable truth the market swallows without questioning: we treat massive AI adoption as a shortcut to brilliance. But it’s actually outsourcing our judgment. When you spend 10 hours a day immersed in logical flows and optimized prompts, your brain changes. Your mind hits a wall with the machine itself and s

CMOs should experiment with AI tools. Broadly, freely, without apology. But they should not be the ones responsible for making those tools actually work. That's a different job. And it's a brutal one. Making AI deliver real results means testing, iterating, throwing tools out, switching methodologies, failing, adjusting, and then doing it all over again next month because every single tool changed while you were figuring out the last one. It's not a skill you develop once. It

Every other day there are posts claiming that Claude is killing copywriters, marketers, developers, and more. The funny thing is, last week I just replaced the Claude Code with the Chinese model GLM 4.7 And speaking of replacement. It's well known that China is playing the geopolitical game to outplace the US. But do you know the big differences between China and the United States in their approach to mass unemployment generated by AI? China is not afraid of AI taking jobs. A

I spent weeks building agentic systems with dozens of AI agents. Neuro-sales, social psychology, Freudian and Lacanian psychoanalysis, ABM, value-based sales frameworks. A full architecture designed to amplify every single touchpoint of my sales process. 80% went in the trash. Not because the technology failed. Because I failed to notice, early enough, when it had stopped amplifying my judgment and started replacing it. AI is not neutral. It's a biased mirror. It amplifies w