Here's how Mike Dee went from spending $1000/m to $25/m for script writing.
​
All with the same effort.
​
In our mastermind call last week and Mike Dee shared some insights about his current script writing process.
​
He went from paying a scriptwriter $1K/month (who still needed editing) to getting 95% ready scripts in minutes.
​
I wrote about this earlier last week. Make sure to follow me on LinkedIn to see these insights earlier.

The AI Script Writing System That's Replacing $1,000/Month Writers

Stop Fighting AI. Start Training It.

Ai is an amazing tool that can help you create better videos and faster but only if you use it in the right way.

Most people type "write me a YouTube script about [topic]" and wonder why the output sounds like every other generic piece of content on the internet.

Mike went from paying a scriptwriter $1,000/month (who still needed heavy editing) to getting 95% ready scripts in 15 minutes by using Claude ($25/month).

All with the same effort for him.

The difference? He stopped asking AI to be creative and started teaching it to be him.

The Context + Brief Framework

The secret isn't in the prompt. It's in the system.

Think of AI like hiring a new writer. You wouldn't just say "write something good" and expect magic. You'd give them:

• Your brand guidelines
• Examples of your best work
• Your voice and tone
• What topics you cover
• What you never would say

Same principle applies here.

The Two-Part System:

Part 1: Context (Build Once, Use Forever)

This is your AI's training manual.

Create a document on Notion or Google Docs with all this context in it. I prefer Google Docs as it's easier to integrate into LLMs.

Include:

Your Speaking Style:

• Pace, energy level, vocabulary choices
• Common phrases and transition words you actually use
• Humor style and personal anecdote frequency
• Technical jargon vs conversational balance

This can be from transcripts from existing videos/keynotes/podcasts/...

Your Content DNA:

• 2-3 of your best handwritten scripts (the ones that sound exactly like you speak)
• Hook styles that have worked for your audience
• Stories and examples you reference frequently
• Topics you avoid or embrace

This part is usually the hardest for (AI) writers to figure out, it's like the everyday stories and anecdotes that make a story and script personal and exclusive.

Your Target Audience:

• What, where and how are they consuming content
• Who are they
• What are their pain points, their struggles

Depending on who you target, your messaging would be different.

Your Performance Data:

• What hooks drove the highest retention
• Which pattern interrupts worked best
• Successful call-to-action approaches
• Video structures that kept people watching

I like to match retention data with the transcript to identify what works and what doesn't.

Part 2: Brief (Unique Per Video)

This is project-specific direction:

• Video topic and target keywords
• Target length (in words, not minutes)
• Main goal and call-to-action
• Key points that MUST be included
  This can be your rough thoughts, bulletpoints etc
• Audience problem you're solving

The Feedback Loop That Makes AI Smarter

Here's where most people stop. They get their script, use it, and move on.

Instead:

1. Use the AI draft (with minor human edits)
2. Publish the video and track performance
3. Feed the results back into your context profile
4. Document what worked (and what didn't)
5. Update your AI training for next time

This creates a compounding effect. Each video makes your AI better at writing like you.

Mike's Bullet Point Method

The fastest implementation I've seen:

Step 1: Create a Google Doc with two sections:

• Requirements (your standard script guidelines, similar to the context)
• Content Brain Dump (bullet points of your raw thoughts, the video brief)

Step 2: In Claude, create a project with:

• Your handwritten script examples
• Clear instructions ("never use 'game-changer,' avoid bullet point overload")
• Your hook and CTA preferences

Step 3: For each video:

• Spend 10-15 minutes dumping bullet points (typos don't matter)
• Copy/paste into Claude with your requirements
• Get back a script that sounds like you

Step 4: Copy to teleprompter and record.

Total time investment: 15-20 minutes vs. hours of manual writing.

The Authentication Balance

One warning: Don't make your AI too authentic.

If you have filler words, grammatical mistakes, or speaking patterns that hurt retention, don't train the AI to copy those.

The goal isn't perfect imitation. It's optimized authenticity.

Your voice, but cleaned up for maximum impact.

Advanced: The Google Docs Integration

Pro tip from our system at Playstack:

Link Google Docs to Claude so when you update your context profile, it automatically reflects in your AI training.

This means as you discover new hooks that work or stories that resonate, you can update once and improve all future scripts.

We're tracking hundreds of scripts across dozens of clients, matching them with retention data to identify what actually drives results.

The pattern we're seeing: More context = more authentic output = better performance.

Your Next Steps

1. Audit your current process - How long does script writing take you right now?
2. Gather your context - Find 2-3 scripts where you nailed your voice
3. Set up your system - Create the context + brief framework
4. Start small - Test with one video and measure the difference
5. Iterate and improve - Feed performance data back into your training

The creators winning with AI aren't using it as a replacement for creativity.

They're using it as an amplifier for their authentic voice.