Oh, AI, you're so agreeable.
AI will just agree to agree to disagree, or disagree to agree to disagree, or whatever.
What am I talking about? Well, AI is notoriously, annoyingly, and even dangerously agreeable.
Just as in our daily life, an overly agreeable "yes-person" can lead us astray and get us into trouble. We often turn to AI to validate a new service offering, build a content calendar, or review the copy on a new sales page. We want it to be an intelligent sounding board.
The problem with an agreeable AI is that its "yes" is often built on false information—or worse, a purely algorithmic desire to please you.
The Dangers
Operationally, this is dangerous. If you pitch a terrible, overly complicated marketing funnel to ChatGPT, it won't tell you it's a waste of your resources. It will say, "Great idea!" and eagerly write the 15-part email sequence for it. You end up burning weeks of your limited time and thousands of dollars launching a campaign that completely alienates your target audience, simply because the AI didn't have the context to tell you "no."
Spotting The False Positives (and Negatives)
The team here at Catalyst recently ran some experiments to analyze a website. We leaned on two different prompts and analyzed the output. Each prompt leveraged the foundational context of the business but lacked guidance context (the guardrails against overly agreeable AI results).
One prompt asked for a brutal teardown of the site, while the other asked for a more general analysis and critique.
The AI gave us equally agreeable responses to both prompts.
The brutal teardown prompt yielded some interesting insights; however, in order to be agreeable to the "brutal" instruction, it was overly aggressive in its negativity. It just became agreeably disagreeable, missing the actual strategic flaws.
The more natural prompt also offered insight but missed opportunities to really help shape the site to meet prospects where they were mentally. Because it wanted to be polite, it allowed terrible messaging and imagery to linger—the exact kind of messaging that turns prospects off and costs the business valuable sales.
The solution? A system that blends guidance through inclusion and exclusion language—detailed, explicit instruction about what your business actually does, what your constraints are, and what success looks like. Sounds like a lot, yeah?
That's because to leverage AI correctly, you absolutely must lift heavy on the front end. If the AI doesn't know your North Star metric, it will just smile, nod, and help you flawlessly execute the wrong strategy.
Us Humans Are Here to Help
The good news? The humans here at Catalyst (yep, real live humans with decades of experience and context) are here to help lighten that lift for you.
Let's start from the top and give your AI the compass it needs to actually help you grow.
Join our free workshop to define your North Star and start creating your perfectly disagreeable AI!
