Mel Robbins is 57. She runs one of the most listened-to podcasts in the world. Her entire production team uses AI every single day. And when she sat down with AI educator Allie K. Miller on her podcast, she admitted something a lot of us feel but don't say out loud: "It has exploded, it has accelerated, and I don't want to get left behind."
That conversation is worth your time. But what I want to pull out of it — the thing that changes everything for a first-time or frustrated AI user — is a simple four-part framework that determines whether your AI results are useful or generic.
If you've already read Context is Everything, you know why your instincts are your biggest advantage here. This is the part where you actually use them.
Why your first prompts probably aren't working
Most people open ChatGPT and type the way they'd type into a search engine. Short. Vague. A few words. And they get back something that's technically an answer but isn't actually useful — which quietly confirms their suspicion that AI isn't really for them.
It is for them. The prompt was just missing four things.
Allie K. Miller put it plainly on the Mel Robbins Podcast: the average person isn't bringing in enough context. They're asking AI to plan a family vacation to Greece without saying who's going, what the budget is, or whether anyone hates museums. The AI isn't holding out on you — it just doesn't know what you didn't tell it.
Here's the fix.
The four-part prompt formula
Every strong prompt has three components. You don't need to use them in a rigid order. You do need all four.
1. Give AI a role
Before you ask it to do anything, tell it who to be. This is the part most people skip — and it's the part that changes the tone, the vocabulary, and the assumptions behind every word of the response.
Without a role, AI defaults to a generic middle ground. With one, it calibrates immediately.
Weak: "Help me plan a trip to Italy."
Strong: "You are an experienced travel advisor for independent women travellers in their 50s who prefer boutique hotels, local restaurants, and avoiding tourist traps. I'm planning 10 days in Italy in September."
Same request. Completely different result.
The role doesn't have to be elaborate. "You are a direct, no-nonsense editor" does the job. "You are a supportive career coach for women returning to the workforce after a break" does even more. Think of it as setting the scene before the conversation starts.
2. Give AI the task
This sounds obvious, but the gap between a vague task and a specific one is where most prompts fall apart.
You're not just giving AI a topic. You're naming the exact output you want.
Weak: "Help me with my LinkedIn bio."
Strong: "Write me three versions of an opening paragraph for my LinkedIn bio. One should lead with experience, one with what I'm looking for next, and one with personality. Keep each under 60 words."
Three versions. Specific format. Word limit included. That's a task. "Help me with my bio" is a conversation starter that puts the work back on the AI to guess what you need.
The more precisely you name what you want, the more useful the response — and the less back-and-forth required.
3. & 4. Add context and constraints
This is where decades of experience become a genuine competitive advantage.
AI doesn't know your situation. You do. The things that would change the answer — who the audience is, what you've already tried, what you want to avoid, what tone matters, how long it needs to be — all of that needs to go in. Not as a paragraph of background, but as specific details that shape the output.
Example: "Write the summary in an email format I can share. Keep it under 150 words. Warm but clear — the reader is my elderly mother and she'll be anxious if it sounds clinical. Don't use medical jargon."
Those aren't restrictions. They're instructions. They're the difference between a response you can use and one you need to rewrite.
Constraints are not a sign that you're making things harder. They're the signal that you know exactly what you need. And knowing that — clearly, specifically, with the human dimension accounted for — is something midlife women have been doing in high-stakes situations for decades.
See the difference
Here's the same request, twice.
Without the formula:
"Help me plan a trip to Portugal."
With the formula:
"You are an experienced travel planner specialising in solo travel for women. I'm 58, travelling alone to Portugal for 10 days in October — my first solo trip. I want a mix of Lisbon, one coastal town, and somewhere quieter inland. I'm a confident walker but don't drive. I prefer small independent restaurants over tourist traps and boutique hotels over big chains. My budget is $5k. Give me a rough day-by-day outline with realistic travel times between locations."
You don't need to see the output to know which prompt gets something useful.
One more move worth knowing
If you're not sure what context to give, ask AI to draw it out of you.
Try this: "I want help with [X]. Ask me the questions you need to give me a great answer."
What comes back will sharpen your thinking and almost always produce a better result than whatever you would have written cold. You've been asking the right questions to get to the real problem your whole life — getting the information out before jumping to solutions. This is the same skill. You're just on the receiving end.
Role. Task. Context. Constraints.
That's the formula. It's not complicated. It's just not what anyone tells you when you first open the tool.
Try it today on something you've been putting off — an email, a decision you need to think through, a plan that's been sitting half-formed in your head. See what comes back when you give AI something real to work with.
Consider it decoded.
Written by Amanda · Midlife TechCurious