AI Prompting That Works: Stop Treating AI Like Google

Most people treat AI like a vending machine. Put in a vague request, get a vague result, complain that AI doesn't work.

I've been managing remote virtual assistants for years. And I'll tell you something: the way most people prompt AI is the same way bad managers brief their team. No context. No clarity. Then they're surprised when the output misses the mark.

AI is a team member. It needs onboarding. It needs your business strategy, your voice, your audience, your pain points. Skip that, and you'll get generic content that sounds like it came from a template factory.

Nat Koko — an ex-Navy tradie turned business brand consultant — ran a TTV workshop on this exact problem. Her take is refreshingly direct: you are the hero, the tool is just a tool.

The Problem With How Most People Prompt

"Write me a post about leadership."

That's not a prompt. That's a Google search.

And when you treat AI like Google, you get Google-level answers. Generic. Safe. Forgettable.

The shift? AI isn't a search engine. It's a skilled assistant who needs context to do their job well. Same as any VA or contractor you've ever worked with. The more context you give, the better the output. Every time.

What is the RTCF Prompting Framework?

Nat shared a framework in the workshop that clicked immediately — because it's basically how I brief my VAs. RTCF stands for Role, Task, Context, Format.

Role: Who do you want the AI to be? An SEO expert? A copywriter? A business strategist with experience in your industry? This matters because it shapes how the AI approaches your request. Without a role, you're asking a generalist to do specialist work.

Task: What specifically do you want it to do? Not "help me with marketing" but "review this landing page for conversion issues." Clarity beats cleverness every time.

Context: This is where most prompts fall short — and where my years of managing VAs taught me the most. What's the background? Who's the audience? What have you already tried? What constraints exist? When I onboard a new VA, I give them my brand voice guide, my customer pain points, my core messaging. Same principle with AI. The more context you provide, the more relevant the output becomes.

Format: How do you want the answer delivered? Bullet points? A structured email? A comparison table? If you don't specify, you're leaving it up to chance.

The Technique That Changes Everything

Here's the part of the workshop that shifted how I think about prompting.

Instead of asking AI to just do the thing, Nat starts by asking it to ask her questions first.

Something like: "Before you give me an analysis, ask me three questions for any clarification you might need."

Why does this work? Because you don't always know what you're missing. And AI, without that clarification step, will just fill in the gaps with assumptions.

When you force it to ask questions, two things happen. You surface gaps in your own thinking. And the AI gets the context it actually needs.

It's slower upfront. But the output is dramatically better.

Set Up the AI Before You Start

One pattern Nat uses is telling AI the outcome before diving into the task.

Instead of: "Here's what I've done, what do you think?"

Try: "Hey, I'm going to be uploading context about [subject] for my business. Before we go any further, I want to set myself up for success with a sharp prompt that's going to get us the outcome we're working towards. Can you provide me a framework so I can prompt effectively?"

You're thinking about the end before you start. That's the difference between prompting reactively and prompting with intention.

Side Pairing: Using AI to Improve Your Prompts

This was a new concept for me.

Side pairing means using one AI chat to improve what you're doing in another.

Say you have a messy conversation going in one chat. You can open a second chat and ask: "How could I have structured this question better to get a clearer outcome?"

Then you take that improved prompt back to your original conversation.

It's like having a thinking partner who helps you communicate more clearly with your other thinking partner.

AI Is Trained to Be Nice (Override It)

Here's something worth knowing: AI is trained to agree with you.

Nat puts it bluntly: "AI is trained to be nice to us. If you just want a no-bullshit approach, you have to train it."

She's done exactly that. Her AI will tell her when she's waited too long to act or when her thinking is off. But that didn't happen by default. She set it up that way.

If you want direct feedback, you need to ask for it explicitly. And then remind the AI to actually deliver it—because its default is to soften everything.

Frequently Asked Questions

Does the RTCF framework work for all AI tools?

Yes. Whether you're using ChatGPT, Claude, or any other tool, the structure works the same. The platform might have different features, but the principle of communicating clearly with a knowledge base to receive a useful answer is universal.

How do I know if AI made something up?

Ask where the information came from. "Where are you referencing this from?" If it can't give you a source, treat it with scepticism. The responsibility for verification stays with you.

Should I worry about getting prompts "perfect"?

No. Better prompting is iterative. Start with something, see what you get, refine. The clarifying questions technique helps because it turns prompting into a conversation rather than a one-shot attempt.

The Bottom Line

AI is a team member. Onboard it properly and it delivers. Skip the context and you'll get slop.

Give it a role. Give it a clear task. Give it the context it needs — your voice, your audience, your strategy. Tell it how to format the answer.

And when you're stuck, ask it to ask you questions first. Same thing you'd want a good VA to do before they start working on something they don't fully understand.

The goal isn't to become a "prompt engineer." It's to get better at articulating what you actually want — which is a useful skill whether you're talking to AI or humans.

Want the full system for creating content that sounds like you? The PACE Mini Course gives you 4 steps to personal branding — including how to use AI without losing your voice.

[Take the PACE Mini Course →][PACE course link]

🍗 String

This post is based on a TTV workshop with Nat Koko, a business brand consultant who helps business owners stop sounding like everyone else and build brands that feel authentic. You can find Nat on LinkedIn.