Last month, a client sent me feedback on a landing page we wrote. It was six pages long, packaged as a PDF file. Clients don't send six-page PDFs.
A week later, a different client did a very similar thing. They copied feedback generated by Claude into a separate tab in a Google doc and asked us to review Claude's recommendations.
That's when I started noticing it everywhere.
In today's newsletter:
A head of marketing on our new project filled in the spreadsheet with client info that I asked for. It was AI-generated, full of fragmented sentences, such as: "Build a GUI on top of ECW's APIs — custom backend orchestrating integrations, proxying Helo to deprecate it over time." During the call, I asked the guy to walk me through the project context and explain what that meant. Well, he couldn't, because he hadn't written any of it. Hell, he didn't even read it.
Here is another example. We had a call with a potential client who had sent me an AI-generated website with brand messaging beforehand. It included positioning and ICP described from every imaginable angle. But when I asked them on the call who their ICP was, they answered: "People who manage projects." That's surprisingly generic coming from someone whose brand messaging is so detailed.
My sister, a UI/UX designer, posted about a design version of this on LinkedIn recently. Product managers ask AI to write a product requirements document, then drop the file into Slack for the designer and developer to sort through. Both spend the next hour figuring out what's a requirement and what's AI filler. The designer's job used to be thinking alongside the PM. Now it's decoding a document nobody on the sending end fully understands either.
The problem I just described comes down to outsourcing thinking to AI.
When you use AI to generate feedback, write specs, or fill in spreadsheets, it creates polished outputs that look finished. Comprehensive, even. But they're so wordy that they create cognitive load and people prefer not to read them. If you don't compress whatever AI has generated, you skip the decisions about what matters and what doesn't. You basically outsource your thinking to AI.
Ironically, in this case, AI doesn't remove work. Every hour someone saves generating a document is spent by someone else trying to interpret it.
This means ↓
When everyone can write, the competitive advantage of anyone trying to answer a question is no longer putting words into sentences. It's understanding them well enough to compress them, challenge them, and decide what matters.
When generating anything with AI, here are a few things I do to avoid “outsourcing my thinking”:
If you manage people who use AI in their output (which is everyone now) the rule worth setting is: they need to be able to say it back in their own words.
Seeing AI-generated text everywhere bothers me far less than people forwarding AI-generated text they haven't even read. Is it just me?
Kateryna
P.S. If we aren't connected already, follow me on LinkedIn and Instagram. If you like this newsletter, please refer your friends.
P.P.S. Need help with quality content? Zmistify your content with Zmist & Copy.


How to use my content scoring system to audit your content and prove the impact of your work.
Subscribe to From Reads to Leads for real-life stories, marketing wisdom, and career advice delivered to your inbox every Friday.