Stop Rewriting Prompts: Use These 4 AI Phrases Instead
AI FOR ORDINARY PEOPLE
The Four Phrases That Turn a Messy AI Chat Into a System
Most people use AI the hard way.
They open Claude, ChatGPT, or another AI tool. They type a messy prompt. They hope it works. Then they rewrite the same instructions again the next day.
In this episode of AI for Ordinary People, Joe Foley breaks down four simple phrases that can turn a messy AI chat into a repeatable system.
These are not magic prompts. They are practical habits.
The four phrases are:
* Interview me
* Run sub agents
* Build the skill
* QA your work
Joe shares how he learned this the hard way while building AI workflows for podcast research, content creation, and everyday work. Instead of retyping long prompts over and over, he started building repeatable systems that save time and create better results.
This episode is for anyone using AI for work, podcasting, writing, research, business ideas, or daily tasks.
You do not need to become a prompt engineer. You just need a better way to start.
IN THIS EPISODE, YOU WILL LEARN
* Why vague prompts lead to weak AI results
* How "Interview me" helps AI ask better questions before it starts working
* Why long single chats can get messy and repetitive
* How sub agents can split research, writing, editing, and planning into separate focused tasks
* Why saving a great result as a reusable skill can save hours later
* How to ask AI to QA its own work before you reuse or scale a process
* Why checking the output matters before turning it into a repeatable system
* What Claude Sonnet 5 means for ordinary users, creators, and small business owners
* Why better AI tools still need human judgment
THE FOUR AI PHRASES FROM THIS EPISODE
1. INTERVIEW ME
Use this when you have a rough idea but do not know how to explain it clearly.
Instead of dumping a long messy paragraph into AI, ask it to interview you first. Let it ask questions. Answer them. Then let it build a clear project brief from your answers.
Example prompt:
"Interview me before you start. Ask me the questions you need so you understand the project clearly."
2. RUN SUB AGENTS
Use this when a project has more than one moving part.
Instead of putting research, writing, editing, and strategy into one long chat, split the work across separate chats. Give each one a clear job.
Example:
One chat researches real questions people are asking
One chat ranks the strongest ideas
One chat drafts the content
One chat edits the final version
This keeps each AI session focused and reduces repeated or confused output.
3. BUILD THE SKILL
Use this after AI gives you a result that works.
Do not trust yourself to remember how you got there. Capture the process. Ask AI to turn the steps into a reusable instruction or skill that you can use again.
Example prompt:
"Turn the steps we just used into a reusable skill I can use next time."
4. QA YOUR WORK
Use this before you save or scale anything.
Ask AI to check its own output against your actual goal. Have it score the result honestly. If the score is below 95, ask it to fix the weak spots.
Example prompt:
"Review this against my goal. Score it from 1 to 100. If it is under 95, tell me what is weak and fix it."
MAIN TAKEAWAY
AI gets better when you stop treating every chat like the first chat.
Ask better questions first. Split complex work into focused parts. Save what works. Check the output before you reuse it.
That is how a messy prompt becomes a system.
TRY THIS THIS WEEK
Pick one task you keep doing from scratch.
It could be writing an email, planning a podcast episode, researching a guest, creating a YouTube description, or preparing for a meeting.
Start with this:
"Interview me before you start."
Then answer the questions.
Once the result works, ask AI to turn the process into a reusable skill. Before you save it, have AI QA the work and fix anything weak.
Small habits compound. Bad systems do too, which is rude but accurate.
LINKS MENTIONED IN THIS EPISODE
Free AI cheat sheet:
https://aiforordinarypeople.com/free
AI for Ordinary People website:
https://aiforordinarypeople.com
AI for Ordinary People on Apple Podcasts:
https://podcasts.apple.com/lb/podcast/ai-for-ordinary-people/id1785498654
Anthropic Claude Sonnet 5 announcement:
https://www.anthropic.com/news/claude-sonnet-5
Claude Sonnet overview:
https://www.anthropic.com/claude/sonnet
TechCrunch article on Claude Sonnet 5:
https://techcrunch.com/2026/06/30/anthropic-launches-claude-sonnet-5-as-a-cheaper-way-to-run-agents/
SUBSCRIBE AND SHARE
If this episode helped, share it with one person who keeps rewriting the same prompt and wondering why AI feels harder than it should.
For more simple, practical AI tips, visit:
https://aiforordinarypeople.com
Become a supporter of this podcast: https://www.spreaker.com/podcast/ai-for-ordinary-people--6447099/support [https://www.spreaker.com/podcast/ai-for-ordinary-people--6447099/support?utm_source=rss&utm_medium=rss&utm_campaign=rss].
If this episode helped you, take a moment to follow the show and share it with someone curious about AI. You can find more simple tools and guides at aiforordinarypeople.com, and you’re always welcome to send in a question or a topic you want covered.
Comments
0Be the first to comment
Sign up now and become a member of the AI for ordinary people community!