The Iterate, Don't Restart Rule

The Problem With Starting Over

When AI gives you something that is almost right, many people delete it all and start from scratch. This wastes time.

New prompt = new attempt = new risk of getting something different that is still not what you want.

The Better Way: Iterate

Iterate means making small changes based on what you got. You keep what works and fix what doesn't.

How to Iterate in 3 Steps

Step 1: Get First Draft

You: "Write an email to customers about our new pricing."

AI: [Email output that is close but not perfect]

Step 2: Tell AI What to Change

You: "The email is good. But make it shorter (1 paragraph instead of 2). Also, remove the part about the old pricing."

AI: [Revised email]

Step 3: Tell AI What to Keep

You: "Good. Keep the tone friendly. Now add a sentence about a free trial offer."

AI: [Another revision]

You are done in 3 steps instead of starting over 3 times.

Why Iteration Works

1. AI remembers the conversation. It knows what you already approved.
2. You are asking for specific changes, not a complete redo.
3. You get to refine the output until it is perfect.

Real Example: The Full Iteration

Step 1: First Draft

You: "Write a cover letter for a job I want. The job is for a project manager at a tech company. I have 5 years of experience."

AI output: [Generic cover letter with decent structure but missing key details]

Step 2: First Iteration

You: "Good start. But make it more specific to THIS company (Acme Tech). Also, add a specific example of a project I managed successfully."

AI output: [Cover letter now mentions Acme Tech but your project example is vague]

Step 3: Second Iteration

You: "Better. The company reference is good. But rewrite the project example to be more impressive. Add: the budget I managed, the team size, and the result."

AI output: [Cover letter with much better project description]

Step 4: Final Polish

You: "Perfect. One last thing: make the opening sentence more personal. Tell them why I specifically want to work at Acme Tech."

AI output: [Final, polished cover letter]

That is 4 interactions to get something great. If you had started over after step 1, you might still be working on it.

Common Iteration Patterns

Pattern 1: Add Missing Information

First draft is too generic.

"Add specific details about [topic]."

Pattern 2: Change One Thing

One part is wrong, the rest is good.

"Keep everything the same, but change the [part] to be [how]."

Pattern 3: Make It Shorter or Longer

Length is off.

"Make it shorter by removing [what] and keeping [what]."

Or:

"Make it longer by adding [what]."

Pattern 4: Tone Adjustment

Tone is not quite right.

"Keep the content, but make the tone more [how]."

When to Start Over

You should restart, not iterate, when:

• AI completely misunderstood what you wanted
• The output is in the wrong format and too different to fix with one prompt
• You realized you wanted something completely different

For everything else, iterate.

The Time Savings

Starting over approach:

1. Prompt 1: Get wrong answer
2. Delete it
3. Prompt 2: Get wrong answer
4. Delete it
5. Prompt 3: Get close to right
6. Delete it
7. Prompt 4: Get right answer

Total: 4 prompts, lots of time wasted

Iteration approach:

1. Prompt 1: Get close answer
2. Prompt 2: Fix one thing
3. Prompt 3: Polish one more thing

Total: 3 prompts, finished faster

Key Takeaway

When AI gives you something close, do not throw it away and start over. Instead, ask for specific changes. You will finish faster and get better results.