Lesson 1.7

Refining Instead of Restarting

Purpose

By the end of this lesson, you will stop deleting AI’s first answer and starting over. You will know how to give feedback, ask follow-ups, and steer a conversation toward exactly what you need - without losing the context you have already built.

Use This When

  • AI’s answer is close but not quite right
  • You are tempted to open a brand new chat and try again
  • The output feels too formal, too long, too short, or slightly off-topic
  • You already gave AI context (your Brand Voice, your business info) and do not want to retype it all

Before You Start

Have one recent AI answer in front of you that missed the mark. You will practice fixing it in this lesson. If you do not have one, ask AI right now: “Write a one-paragraph welcome email for my business.” We will fix whatever it gives you.

Conversation Prompt

Copy and paste this into your AI chat: “The answer you just gave me is a starting point, not the finish line. I am going to give you feedback in plain English, and I want you to revise based on that feedback. Do not apologize or explain what you did wrong. Just fix it and show me the new version.

Here is my feedback:”

Then fill in one or more of these:

  • “Make it shorter by about half.”
  • “Make it sound more like my Brand Voice Profile.” (Paste your profile here if you have not already.)
  • “Lead with the benefit to the customer, not the feature.”
  • “Remove the phrases: [list words you hate].”
  • “Add a specific example about [your product/service].”
  • “Change the tone from formal to conversational.”
  • “This is good, but the ending is weak. Give me three stronger options.”

Dobie Tip

AI does not get its feelings hurt. You are not being rude by saying “That is not what I meant.” You are being efficient. The magic is not in the first prompt. The magic is in the fourth or fifth follow-up when the answer finally clicks.

Follow-Up Questions

  • “That is closer, but still too wordy. Cut 30 percent without losing the meaning.”
  • “The middle paragraph is perfect. Keep that. Rewrite only the beginning and end.”
  • “Imagine you are talking to a tired parent at 9 PM, not a CEO at 9 AM. Adjust.”

  • “What would you need to know from me to make this truly excellent instead of just acceptable?”
  • “Give me two versions: one safe and one bold. I will tell you which direction I like.”

Expected Result

By the end of this lesson, you should have:

  • A revised piece of content that is noticeably better than the first draft
  • The habit of treating AI’s first answer as a rough draft, not a final product
  • A few go-to phrases for redirecting AI without starting over
  • Confidence that you can salvage a “bad” answer instead of abandoning it

Common Mistakes

Avoid these refinement traps:

  • Deleting the chat and starting from scratch because the first answer was wrong. (You lose all your context. Fix it here instead.)
  • Saying “make it better” without naming what is wrong. (AI is not a mind reader. Point to the specific sentence or tone that missed.)
  • Accepting the second draft if it is only slightly better. (Push one more round. The third draft is often where it lands.)
  • Asking for a rewrite without telling AI what to keep. (If part of the answer is good, say so. Otherwise, AI throws out the baby with the bathwater.)

Instead, remember:

  • Your first prompt gets you in the ballpark. Your follow-ups get you to home plate.
  • “That is close” is a perfectly valid instruction. It tells AI to adjust, not restart.
  • AI has no memory of the “bad” version once you move on. It only cares about the current conversation.

Why This Matters

Most people quit on AI too early. They get one generic answer, assume AI is overrated, and go back to doing it themselves. The people who get incredible results are the ones who stay in the conversation for three or four rounds. They treat AI like a collaborator, not a vending machine. Refining is where the real work happens - and it is still ten times faster than writing from scratch.

Save This

Save this feedback template somewhere you can paste it quickly: “Revise this based on my feedback:

  • Keep: [what works]
  • Change: [what is wrong]
  • Tone: [adjustment needed]
  • Length: [shorter/longer/specific word count]
  • Do not: [anything to avoid] Show only the revised version.”

Use it every time AI’s answer is 70% there. It will save you from starting over.

Coming Up Next

Lesson 1.8: Working with Different AI Models. Not all AI tools work the same way. Learn when to use which one - and why copying the same prompt into ChatGPT, Claude, and Gemini can give you wildly different (and useful) results.