Artificial Idea | AI careers · practical prompts · no hype Thursday, August 7, 2025 · Issue #2 · Prompt Tutorial
Your first 5 prompts: a beginner's starter kit for non-technical professionals
You don't need to understand how AI works to use it well. You need five prompts, a browser tab, and twenty minutes. Here they are.
On Monday we talked about the bifurcation happening inside the labour market right now , the widening gap between professionals who are integrating AI into their daily work and those who are watching from the sideline. The data on that gap is unambiguous. What is less clear, for most people, is where to actually start.
The barrier is rarely motivation. Most professionals reading this understand, at some level, that learning to use AI tools is no longer optional. The barrier is entry. The tools feel technical. The vocabulary , prompts, models, tokens, context windows , sounds like it belongs to a different industry. And the advice available online tends to assume either that you know nothing at all, or that you are already building automations in a coding environment.
This newsletter exists in the space between those two extremes. You are a professional. You have real work to do. You need AI to make that work faster, sharper, and less exhausting , and you need to get there without a technical degree or a week-long course.
So here is where you start. Five prompts. Each one solves a specific, recurring problem in a knowledge worker's week. Each one works in ChatGPT, Claude, or Gemini , open a free account in any of them and you are ready.
First, one thing to understand about how prompts work
A prompt is not a search query. When you type something into Google, you are trying to match keywords to existing content. When you write a prompt, you are giving instructions to something that reasons , imperfectly, but genuinely , about what you need.
The quality of what you get back is almost entirely determined by the quality of what you put in. Vague prompt, vague output. Specific prompt, specific output. This is the single most important thing to internalise before you write your first one, because it explains why most people's first experience with AI is underwhelming. They typed something short and general, got something short and general back, and concluded the tool wasn't useful.
The tool is useful. The prompt needed more work.
With that said , here are five that already have the work done for you.
Prompt 1: The email drafter
The problem it solves: You need to write a professional email that is sensitive, clear, or just time-consuming to get right , a follow-up after a difficult meeting, a request to a senior stakeholder, a rejection that needs to stay warm.
You are a professional communication coach helping me write a work email.
Context: [describe the situation in 2–3 sentences , who you're writing to,
what happened, what you need them to do or understand]
Tone: [professional / warm / direct / diplomatic , pick one]
My key points: [list 2–3 things the email must communicate]
Write a complete email. Subject line included. Keep it under 200 words
unless the situation requires more. No filler phrases like
"I hope this email finds you well."
Why it works: You are giving the model a role, context, tone, and constraints. Each of those four inputs narrows the output from generic to specific. Remove any one of them and the quality drops noticeably.
Prompt 2: The meeting summariser
The problem it solves: You have a page of rough notes from a meeting and need to turn them into something a colleague who wasn't there can actually use.
I'm going to paste my raw notes from a meeting.
Please turn them into a structured summary with three sections:
1. Key decisions made
2. Action items (with owner names if mentioned)
3. Open questions that still need resolution
Keep the language professional but plain. If something in my notes is
unclear, flag it with [UNCLEAR] rather than guessing.
Here are my notes:
[paste your notes]
Why it works: The three-section structure forces the model to organise rather than just reword. The instruction to flag unclear items rather than invent answers is important , without it, AI will fill gaps confidently and incorrectly.
Prompt 3: The thinking partner
The problem it solves: You have a decision to make, a problem to work through, or an idea you want to stress-test , and you need a sounding board that doesn't have a stake in the outcome.
I need help thinking through a decision/problem. Don't give me a
recommendation immediately. Instead, ask me three clarifying questions
first , the ones you'd need answered to give genuinely useful advice.
After I answer, then give me your analysis. Include: the strongest
argument for each option, the biggest risk I might be underweighting,
and what you would want to know more about before deciding.
Here's the situation: [describe it in plain language]
Why it works: Most people use AI like a vending machine , input problem, expect solution. This prompt turns it into a conversation. The clarifying questions step alone tends to surface things the person hadn't fully articulated, which is often more valuable than the answer itself.
Prompt 4: The document explainer
The problem it solves: You have received a long report, policy document, contract, or research paper that you need to understand quickly without reading every word.
I'm going to paste a document. Please do the following:
1. Summarise the main argument or purpose in 3 sentences
2. Pull out the 5 most important points I need to know
3. Flag anything that requires action, a decision, or follow-up from me
4. If there is anything ambiguous or potentially problematic, note it
separately at the end
Here is the document:
[paste document]
Why it works: The numbered instruction set gives the model a clear output structure to follow. Point 3 , flagging action items , is the one most people forget to include and the one that saves the most time.
Prompt 5: The first draft generator
The problem it solves: You need to produce a piece of written work , a report section, a proposal introduction, a client update, an internal briefing , and the blank page is the hardest part.
Write a first draft of [type of document] for the following situation:
Audience: [who will read this , their role, what they care about]
Purpose: [what this document needs to achieve]
Key information to include: [bullet the main points]
Tone: [formal / conversational / persuasive / neutral]
Length: approximately [X] words
This is a first draft for me to edit, not a final version. Prioritise
clarity and structure over polish.
Why it works: The "first draft for me to edit" instruction matters more than it looks. It calibrates the model's output toward structure and completeness rather than surface polish , which is what you actually need at the drafting stage. You bring the polish. AI brings the scaffold.
How to use these this week
Pick one. Not all five , one. Find the problem in your working week that costs you the most time or causes the most friction, match it to the prompt above, and run it. Spend ten minutes adjusting the inputs until the output is genuinely useful.
Then next week, add a second.
The professionals who get the most out of AI tools are not the ones who tried everything at once. They are the ones who got one prompt working well, internalised why it worked, and built from there. That pattern of learning compounds quickly. Trying to learn five things simultaneously compounds nothing.
On Monday we are looking at something the data makes clear but most people find uncomfortable to hear: which specific professions are actually at risk, which ones are not, and why the answer is more nuanced than any headline you have read on the subject.
It is worth showing up for.
The Artificial Idea team
