Generic AI output isn't a tool problem. Halifax AI Integration Strategist Carol Roderick,\ PhD explains the four-part communication skill, borrowed from relationship coaching, that fixes prompting for small business owners and AI communication for good.
"I'm not a tech person."
People say that to explain why AI doesn't work for them. It's the wrong diagnosis.
Or they say this one. "It gave me something clean and completely off. Not wrong. Just not me." Then they rewrite the whole thing anyway and wonder why they bothered asking AI at all.
Before any of this AI work, I trained under Alison Armstrong to become a PAX certified dating and relationship coach. Different career. Same skill, it turns out. It's the skill behind AI adoption for small business owners and consultants who want to use AI without losing their voice in the process.
Treat AI like a partner
People treat AI like a partner who should already know what they need. They hint. They give half the picture and expect the rest to show up right.
Then the output comes back generic, or wrong, and they blame the tool. Or themselves. "Not technical enough."
Something's missing. They're just wrong about what.
AK's read every bland LinkedIn post ever written. Left alone, it writes like all of them. Not a technical gap. The mind-reading gap. You expect someone to guess right. They guess wrong.
Make a "Great Ask"
Alison Armstrong spent years teaching this to couples. I sat through the teleclass notes again recently and realized I'd been half-doing this with AI the whole time, the same way I used to half-do it in my relationships before I learned better.
A great ask has four parts.
I need ___. A plain statement. Not a hint. Not "it would be nice if." Just the need, named. With AI, this is "I need a LinkedIn post" or "I need this rewritten for a Nova Scotia audience."
This would look like ___. The specifics. What, how long, in what tone, using what example. This is the part almost everyone skips. They assume it's obvious. A partner won't guess that right. A language model that's read everything and experienced nothing definitely won't.
This would provide ___. Why it matters. What it's for. Tell AI the post is going to a cold audience who's never heard of you, and it writes differently than if you tell it the post is for existing clients.
What do you need from me to give me this? The question almost nobody asks. With a partner, this is what turns a demand into a partnership. With AI, it's the question that gets you a follow-up instead of a guess. Ask it. It will tell you what's missing.
Catch when you complain about AI
Someone gets AI output that's flat, and instead of making a clearer ask, they complain about it. To themselves, to a colleague, sometimes right back into the chat. "This isn't right." "Too generic." "Doesn't sound like me."
A complaint is a great ask that hasn't been said yet. Take the exact thing that's bothering you and turn it into the four parts above. "This sounds generic" becomes "I need this in my voice, here's what that sounds like, here's a sample, tell me what you need from me to match it."
Nobody taught most people a process for prompting AI well. Copy the email into the chat window, ask for a reply, copy the reply back out, send it. It works. It's also exhausting, and it never gets better, because nothing about the ask improved.
I see it in client work constantly. Same AI, same account, wildly different output. Never the tool. Whether the ask was specific, or just "write me a post" and hope.
Test it in AI
Next time AI gives you something flat, don't rewrite it and move on. Go back to the four parts. What do you need. What would that look like. What would it provide. What does AI need from you to get there.
I used to sit across from couples doing this exact thing badly for years. Still catches me off guard how well it works on a chatbot.
Ready to stop rewriting everything AI gives you? Visit carolroderick.ca and we'll build a "Great Ask" into how you actually run AI in your business.