TL;DR
- Disappointing AI output usually traces to missing context in the prompt, since people leave out tone, format, goals and audience while assuming those things are obvious.
- PAST is a four-part prompt structure covering Problem (the real job to be done), Action (the specific thing to produce), Structure (what the output should look like) and Tone (how it should sound).
- A before-and-after pair shows the difference: “write me a blog post about productivity” against a prompt naming the role, word count, section layout and register, with a ready-made example to paste and try.
- PAST is described as a reliable structure rather than an academic breakthrough, and the post carries a licensing notice stating that PAST and SHAPE are proprietary and commercial use requires licensing.
In plain English
A prompt is the instruction you type to an AI model, and the model works only with what that instruction contains. PAST is a checklist for filling the usual gaps. Problem states the actual job, such as explaining the value of documenting AI workflows to a non-technical stakeholder, which gives the model a target. Action names the specific output with a direct verb, like writing a four-paragraph post or suggesting three headline improvements. Structure fixes the format, such as a table of pros and cons or bulleted points with bold headers, which saves reformatting later. Tone sets the register, for example conversational and slightly witty, or British English, polite and direct.
Iteration matters too. When a response lands at roughly 80% of what you wanted, follow-up instructions like “make this more conversational”, “add specific examples” or “shorten this to 200 words” build on the existing answer instead of starting again.
Signal Over Noise #06
June 13th, 2025
This Week: A Practical Framework for Writing Better AI Prompts
Dear Reader,
Most AI prompts fail before they even begin. Not because you don’t know what you want—but because the AI doesn’t.
When people say “It gave me something weird” or “That’s not quite what I meant,” what they’re often running into isn’t a model problem. It’s a context problem.
Enter: The PAST Framework
This is something I developed not because I needed something new, but because I needed something reliable.
PAST is a simple way to make your prompts make sense—to you, and to the AI.
The Problem With Most Prompts
We humans tend to shortcut instructions. We assume tone, format, goals, and audience are obvious. They’re not—not by a long shot. AI only works with what you tell it, and most people leave out the key ingredients.
PAST solves that by helping you structure your request with four clear components:
The PAST Breakdown
P: Problem
What’s the real job to be done?
You’re not just “writing a post.” You’re trying to explain a specific thing to a specific audience. Define that clearly and the AI has a target.
“I need help explaining the value of documenting AI workflows to a non-technical stakeholder.”
Clearly defining the problem gives the model an understanding of the real issue you’re trying to solve.
A: Action
What do you want it to do - specifically?
“Write a 4-paragraph blog post.”
“Suggest 3 improvements to this headline.”
“Create a checklist for onboarding a new client.”
Vague prompts = vague results. Direct verbs are your friend.
S: Structure
What should the output look like?
“Return it as a table with pros/cons.”
“Use bullet points with bolded headers.”
“Write it in the style of a Twitter thread.”
This step saves you the time of reformatting a wall of text later.
T: Tone
How should it sound?
“Keep it conversational and slightly witty.”
“Use British English. Be polite but direct.”
“Speak like a peer, not a teacher.”
This is where you inject personality, branding, or clarity of voice.
Why I Built PAST
I work with AI every day—building prompts, training agents, automating workflows. And no matter the use case, most of the issues I see come down to one thing: a lack of context.
PAST is my way of fixing that. It’s become the default in how I write, how I build for clients, and how I train others to use AI effectively.
See PAST in Action: Before & After
Here’s what happens when you apply the framework:
Before (typical prompt): > “Write me a blog post about productivity”
After (using PAST): > You are a productivity coach who helps remote workers stay focused. Write a 500-word blog post about morning routines that boost productivity. Structure it with an intro, 3 main tips with examples, and a conclusion. Keep the tone practical and encouraging—like advice from a helpful colleague.
The difference? The first gives you generic fluff. The second gives you exactly what you need.
Try This Prompt Right Now
Copy/paste this into ChatGPT and see the difference:
You are a marketing consultant who helps independent educators build their first online course. Create a short email encouraging someone to turn their blog series into a paid course. Use a 3-paragraph structure. Keep the tone encouraging, practical, and lightly persuasive.
That’s one clean, complete prompt—and it works.
Iterate Like a Pro
Got a response that’s 80% right but not quite there? Don’t start over. Build on it:
“Make this more conversational” or “Add specific examples” or “Shorten this to 200 words”
Think of it as having a conversation with a very capable assistant who needs clear direction. The better your guidance, the better your results.
Coming Next Week: SHAPE
PAST gives you structure, next week, I’ll introduce a second framework for shaping how your AI behaves, responds, and handles nuance. When you combine both, you stop writing prompts… and start building AI teammates.
Want Help Applying This to Your Work?
If you’re still getting “meh” responses from ChatGPT and want to actually use it in your workflows, systems, or content creation—let’s fix that.
Book a 1:1 session with me and I’ll show you how to tailor prompts and AI tools to your real-world needs.
Until next time,
Jim
Signal Over Noise is written by Jim Christian. Subscribe at newsletter.jimchristian.net.
Ready for AI That Actually Works Together?
Stop switching between disconnected AI tools. In a 90-minute AI Action Plan Session, I’ll show you how to set up the kind of orchestrated workflows I use daily—where your AI can read your files, update your systems, and execute complex tasks across all your tools. Let’s design your unified AI workflow.
Book Your AI Action Plan Session Here
Framework Licensing Notice: The PAST and SHAPE frameworks mentioned in this newsletter are proprietary methodologies. While you’re welcome to use these concepts for personal learning, commercial use requires licensing. If you’re interested in team or enterprise licensing, reply to this email.
Made with ❤️ in Valencia by Jim Christian. For feedback, please reach out to [email protected].
