Here’s a pattern I see constantly:
- Someone asks AI a question
- They get a decent-but-not-great response
- They think “AI isn’t that useful” and give up
This drives me crazy. Because they stopped one step before AI gets actually good.
The first response is a draft. The magic happens in the follow-up.
Why First Responses Are Always Mid
When you ask AI something, it’s making a lot of guesses:
- What format do you want?
- How detailed should it be?
- What’s your expertise level?
- What’s the actual goal?
- What tone fits?
Without enough context, AI hedges. It gives you the safest, most generic version of an answer — something that’s “good enough” for most people but perfect for no one.
That’s not a flaw. That’s the starting point.
The Follow-Up Mindset
Think of AI like a first draft from a smart intern. The draft has good bones, but it needs direction.
You wouldn’t throw away a draft because it’s not perfect. You’d mark it up:
- “This section is too long”
- “Can you add an example here?”
- “The tone is too formal — make it friendlier”
- “I need this to be more specific to my situation”
That’s exactly what follow-up prompts are. And they work shockingly well.
10 Follow-Up Prompts That Fix Almost Anything
Bookmark these. They’ll become the most-used phrases in your AI toolkit.
1. “Make it shorter”
The single most useful follow-up. AI tends to overexplain. This fixes it.
Variations:
- “Cut this in half”
- “Give me the TL;DR version”
- “Summarize in 3 bullet points”
2. “Make it more specific”
When the response is too generic or theoretical.
Variations:
- “Give me a concrete example”
- “Apply this to [my specific situation]”
- “What would this look like in practice?”
3. “Make it more casual”
When it sounds like a corporate memo.
Variations:
- “Write this like you’re texting a friend”
- “Less formal, more human”
- “Pretend you’re explaining this at a bar”
4. “Make it more actionable”
When you get theory but need steps.
Variations:
- “What should I do first?”
- “Give me a step-by-step checklist”
- “What’s the smallest action I can take right now?”
5. “What am I missing?”
When you want AI to challenge your thinking.
Variations:
- “Play devil’s advocate”
- “What’s the counterargument?”
- “What could go wrong with this?”
6. “Explain it differently”
When the explanation isn’t clicking.
Variations:
- “Use an analogy”
- “Explain like I’m 12”
- “Give me a real-world example”
7. “Focus on [specific aspect]”
When the response is too broad.
Variations:
- “Just the pricing section”
- “Only the technical requirements”
- “Focus on the emotional benefits”
8. “What would [expert] say?”
When you want a specific perspective.
Variations:
- “How would a CFO respond to this?”
- “What would a skeptical customer ask?”
- “Give me the startup founder perspective”
9. “Now make it better”
Yes, this actually works. AI will genuinely improve it.
Variations:
- “Polish this”
- “Take another pass and improve it”
- “What’s the better version of this?”
10. “Combine these into something new”
When you’ve generated multiple options.
Variations:
- “Take the best parts of options 1 and 3”
- “Merge these approaches”
- “Create a hybrid version”
Real Example: Writing a Cold Email
Let me show you how follow-ups transform a mediocre first response into something actually good.
Prompt 1 (First Try)
Write a cold email to pitch my freelance design services to startups.
Response 1 (Generic)
Subject: Professional Design Services for Your Startup
Dear Hiring Manager,
I am a professional graphic designer with 5 years of experience helping businesses create compelling visual identities. I specialize in brand design, UI/UX, and marketing materials.
I would love the opportunity to discuss how my skills could benefit your company. Please let me know if you would be available for a brief call.
Best regards, [Name]
This is… fine. But it sounds like every other cold email ever sent. Let’s fix it.
Follow-Up 1
Make this more casual and specific. I’m reaching out to early-stage SaaS startups who just raised seed funding. Mention that I saw their funding announcement.
Response 2 (Better)
Subject: Congrats on the funding — quick design question
Hey [Name],
Saw you just closed your seed round — congrats! That first-hire chaos is real.
Quick question: do you have design help lined up? I work with early-stage SaaS teams on product design and brand identity, usually in the “we just raised and need to look legit” phase.
Would love to show you a few examples if you’re interested. No pressure either way.
[Name]
Way better. But let’s push further.
Follow-Up 2
Make the subject line more intriguing. And add a specific result from a past client.
Response 3 (Good)
Subject: The thing most seed-stage startups skip (that costs them later)
Hey [Name],
Saw you just closed your seed round — congrats!
Quick heads up from someone who’s worked with 20+ startups at your stage: most teams deprioritize design, then spend 6 months “fixing” their brand after launch.
I help early-stage SaaS teams nail their product design and brand from day one. Last client went from “our app looks like a prototype” to getting featured by Product Hunt — mostly because we fixed the first impression.
Would love to show you a few examples if you’re open to it.
[Name]
Three prompts. Same AI. Completely different email.
The first version would get deleted. The third version gets replies.
The 3-Round Rule
Here’s a simple rule I follow:
Round 1: Get the raw material out Round 2: Fix the biggest problem (usually length or tone) Round 3: Add specificity and polish
Three rounds is usually enough. More than five rounds means you should probably rethink the original prompt.
When to Start Over vs. Keep Iterating
Keep iterating when:
- The structure is right but the details are off
- You need to adjust tone, length, or specificity
- You want to combine elements from multiple responses
Start over when:
- The AI misunderstood the fundamental task
- You realize your original prompt was wrong
- The direction is completely off
If you’re on follow-up #7 and still not happy, the first prompt was probably the problem.
Advanced Technique: The Feedback Sandwich
This is my favorite trick for getting exactly what I want.
After AI gives a response, say:
“Here’s what I like about this: [specific elements]. Here’s what needs to change: [specific issues]. Give me another version.”
Example:
“I like the casual tone and the specific result you mentioned. But the subject line is too long and the opening feels too salesy. Give me another version that opens with a question instead.”
This feedback sandwich gives AI clear direction on what to keep and what to fix. The next version is almost always better.
Why This Matters More Than You Think
Most people treat AI like a vending machine: insert prompt, receive answer.
But the best results come from treating it like a collaboration. You’re directing, refining, pushing back. The AI is generating, adjusting, trying again.
This shift — from one-shot prompting to iterative dialogue — is the difference between “AI is okay I guess” and “AI changed how I work.”
The Meta-Skill
Here’s the thing: once you internalize follow-up prompting, every skill becomes more powerful.
That email writing skill? Use it to generate a draft, then follow up with “make it half as long.”
That code review skill? Ask it to “focus only on security issues in the authentication flow.”
That brainstorming skill? Tell it “those are too obvious — give me weirder ideas.”
The follow-up is where you make generic tools specific to you.
Try It Right Now
Open ChatGPT or Claude. Ask it anything — write an email, explain a concept, whatever.
When you get the response, don’t accept it. Don’t close the tab.
Instead, say: “Make it shorter and more specific to [your situation].”
See what happens.
One follow-up. That’s all it takes to go from “meh” to “actually useful.”
Want more techniques like this? Check out our System Prompt Architect to create custom AI assistants, or browse all AI skills for ready-to-use prompts.