Pick my brain!
Fill out a few details so I can prepare before we chat.
AI writes the copy, builds the pages and optimises the bids. So what is your marketing team for?
Machines write ad copy that converts. They build landing pages overnight, segment audiences, optimise bids, shoot product photography without a studio and draft email flows in the time it takes you to brief a copywriter.
Behind this sits $185bn of infrastructure spending and 160,000 developers building what the industry calls digital employees. Not assistants. Employees. One company now runs a software factory where the code writes, tests and deploys itself.
Shopify Magic writes your product descriptions. Klaviyo builds your segments. Your ad platforms optimise their own creative. Most of what a junior marketer did three years ago, a well-configured agent does now — faster, cheaper and without taking Fridays off.
The gap is in adoption. 71% of firms say they use AI agents; 11% have them in production. That gap is closing. When it shuts, the brands that treated AI as a novelty will find they have been lapped by the ones that treated it as plumbing.
Six things make a piece of marketing work difficult. Machines have taken four of them.
Machines own the first four. They do not tire, they do not get bored and they do not miss the detail in row 47,000 of the customer export.
Novelty and judgment are still ours. In marketing those two decide whether a brand sounds like itself or like everyone else's ChatGPT output.
When execution is automated, four human skills become the whole game. The frame is borrowed from Nate's Notebook.
Knowing that this subject line will land and that one will not. Knowing the difference between a page that converts and one that merely looks professional. AI can write a hundred options. Someone has to pick the right one.
The unwritten rules of your market. Your customers behave differently in January than in March. One segment will hate this campaign even though the numbers say they should love it. Years in a category buy that; nothing else does.
When the ground shifts every quarter, what you know matters less than how fast you learn the next thing. Treat each new capability as leverage rather than threat. Those who learn fastest compound the lead.
Seeing what is happening rather than what you hoped would happen. The campaign is not working. The creative is mediocre. AI tells you what you want to hear. Someone has to say this isn't good enough.
Building used to be the hard part: coding the page, designing the template, writing the descriptions, running the media plan. That is now cheap, and getting cheaper. The new bottleneck is saying precisely what to build.
Marketing teams were built for throughput. You hired people who could do things — design, write, code, run campaigns. When doing is free, the constraint moves upstream, to the people who can define the right thing to build and write a brief an agent can execute against.
Describing what you want requires knowing what you want, which is harder than it sounds. It demands clarity about your brand, your customer and your position. The woolly brief a talented designer used to read charitably? An agent takes it literally, and ships it a thousand times.
In the teams getting this right, AI does the execution: four-fifths of the work. Humans write the specification at the front and judge the result at the end. A fifth of the hours, and it decides whether the other four-fifths were worth doing.
The point in humans is not doing. The point is knowing.
Knowing which campaign to run. Knowing which segment is being underserved. Knowing when the data is lying to you. Knowing the brand voice has drifted. Knowing this creative is right and that one misses by a mile.
The window is short. Brands that build this in the next year or so will have better systems, sharper judgment and faster learning cycles than the ones still using AI as a fancy autocomplete.
The point in humans was never our ability to execute. It was our ability to know what is worth executing. We are just being made to prove it.
The shift from execution to specification is under way. You either lead it or react to it.