I Used AI to Write My Entire $533K Launch — Here Are the Exact Prompts

Last month I launched our AI Assisted Agency program with three YouTube videos and five emails to my list. The result? Over $500,000 in revenue — with almost zero acquisition cost. Those three videos still collectively have only around 13,000 views.

What most people don't know: AI wrote almost everything. The video scripts, the emails, the sales page, the VSSL, the automated email flows — all of it was 90% AI-generated. My job was to edit the scripts, which took a couple minutes each.

Here's exactly how I did it.

The Platform I Used

I used Manis — a credit-based AI platform. I chose it specifically because you can upload a massive amount of context and it actually reads it. That matters more than most people realize. The quality of your output is a direct function of the quality of context you feed in.

Step 1: Write One Email Yourself First

I started by writing a single rough email — basically a freewrite about the offer idea. AI Assisted Agency. Not an AI automation agency like everyone else was building. Just using AI as a fulfillment tool inside a normal service-based agency. Better opportunity, cleaner offer.

Once I had that email, I asked AI: "What are the questions people are probably going to have when they see this offer?"

It came back with: technical implementation questions, business model and pricing concerns, market competition fears, practical execution questions, results and proof questions.

That list is dead accurate. That's literally what people ask me. I used those questions to guide the rest of the email sequence.

Step 2: Let AI Write the Other Emails

I told it: "Review all the emails I've written. Help me write the fourth email. Write it in my style. Sound like me."

It wrote an email addressing common objections: I'm bad at sales. I don't have time. What if clients find out I'm using AI? Exactly what needed to be said.

Step 3: Map Out a Three-Video Launch Series

Here's the prompt that did the heavy lifting:

"I have a YouTube channel with 30,000 subscribers. I want to make a three-video series for the launch of this offer. Imagine yourself to be a direct response copywriter capable of delivering airtight sales arguments that logically lead the viewer to the conclusion that they need to act. The videos should be information-dense and logic-dense, relatable to the fears of anyone running an agency who isn't seeing success. Before writing the scripts, map out the topics, titles, packaging, and what each video is trying to accomplish — and make sure they logically lead into each other."

It came back with a clear strategy:

• Video 1: Create massive problem awareness and fear of being left behind

• Video 2: Reveal the AI mechanism, build desire through proof

• Video 3: Provide the blueprint and bridge naturally to the program

I approved the strategy, then told it to fully script all three. When they came back too short, I said: "Make each video three times longer. Introduce new angles and arguments that strengthen the sales case."

That's how you get a 35–40 minute video script. That's the video that drove the launch.

One Critical Edit I Had to Make

Video two came back with fabricated data — fake case studies, made-up stats. I caught it and told it: "Rewrite this. Don't use any numbers or data you invented." Then I manually inserted my real numbers. That's the extent of the intervention required.

Step 4: Build the Sales Page

I took all the email and video material, threw it into a rough page draft — zero structure, pure freewrite — and told the AI:

"Imagine you're one of the highest-earning direct response copywriters alive, specifically in the business opportunity and information marketing space. Critique this page and give me suggestions to significantly increase bookings."

It broke down everything wrong with it. Then I said: "Now rewrite the entire page implementing everything you just said. Here are the real case studies. Don't invent stories about them."

It rewrote the full sales page. What's live now is basically what it produced.

I also had it generate a website wireframe so I could hand a designer a visual reference. Prompts can do that too.

Step 5: Write the Full Email Funnel

After the sales page, I needed an opt-in sequence. I wrote email one myself, then told AI: "Come up with a strategy for the next five emails. Don't write them yet — just explain the strategy."

Once I approved the strategy, I said: "Write them all out fully." Done.

I also used it to write abandoned application sequences — emails targeting people who filled out an application but never booked a call. I asked it why someone might do that (sticker shock, commitment fear, analysis paralysis), then had it write recovery emails timed at 3 hours, 24 hours, and 48 hours post-abandon.

What This Actually Proves

This entire launch — scripts, funnel, emails, sales page — took me maybe 5 to 7 hours of total work. Before AI, that's easily a 100-hour job.

You don't need to build AI tools or apps. You don't need to be technical. You just need to know how to use AI to produce marketing assets — and then sell that skill as a service to businesses that need it.

High-ticket clients are sitting on dead email lists, dormant audiences, and cold traffic they don't know how to convert. You go in, use AI to script their assets, and take a percentage of what it makes them. If it doesn't work, they don't pay. Zero risk to them.

The only thing standing between you and that is the confidence that you can actually get results. And as I just showed you — AI handles that part.