Mel's AI Journey And Staggering Early Results

Hello Everyone,

I am beyond excited and in awe of what AI is doing for me right now and going forward every future hour.

The last 48 hours have blown my mind.

First, I'll back up a little.

I don't even remember how I got started on this, but using the Command Prompt on my PC, I created a program that goes to Gutenberg and is going through their entire 76,000 public domain titles. It downloads them to itself, counts the words and continually reports its progress.

For it a page represents 32 books it is scanning. When you see it is regarding matches, those are titles with 30,000 words or less which would represent a short book of 200 pages or less.

As of right now that's 2, 678 titles and counting.

What it then does is it creates a CSV file providing me the following info.

Author, Title, Word Count & Text Link to the title online.............Priceless.

Now my Mind Blowing past 24 hours.

Codex is a cloud-based AI software engineering agent integrated into ChatGPT (powered by codex-1/o3 models) designed to automate coding tasks like writing features, debugging, and creating pull requests. It operates in a secure, sandboxed environment, supports repository analysis, and is available to Plus, Pro, and Enterprise users.

This was just introduced to ChatGPT and I wanted to create AI Agents so I dived in. I am on ChatGPT Plus.

I wanted to create a system where I can better use OCR Scans of public domain titles, Have it produce the actual text, clean it up so with minimal editing I can then publish the title.

That was a long 24 hours and lots of tweaking.

Because Codex can cost additional fees, after I gave to Codex what I wanted and it responded, I took that info to ChatGPT and then we fine tuned it to later give back to Codex and complete.

ChatGPT literally had to hold my hand every step of the way. Way above my pay grade or knowledge or technical ability. Despite that, I pushed forward.

So now I can provide the link to an OCR Scanned title and it will go to work instantly and in minutes provide me what I need. WOW. Let me repeat that. WOW!

The next 24 hours, I wanted a system to take a text file and have it cleaned up so it was ready for my final edits to be published.

Oftentimes text for titles can be somewhat of a mess. Hyphenated words at the end of sentences or just plain words broken in half. Paragraphs packed together. Unnecessary Headers and Footers and lot of other things that I don't need to publish the title and its very time consuming to edit that myself.

Codex, ChatGPT Plus and I achieved that.

So here again, I provide the title link in a folder and it immediately goes to work and in MINUTES is finished and I go to another folder to retrieve the results.

So finally it was time to test it with a 350 page title. Boom!

I then copied it and pasted it into Adobe Indesign which is what I use and I am working on my final clean up and editing, making sure it is ready to be published later today.

All I can say is, "If You Can Think It, AI Can Create It".

Since I still need to have my hand held and we have covered so much territory then today I am going to ask for a script for how to do the above to processes when I submit links and for it to instruct me as if I were a 10 year old to keep it simple and understandable.

My ChatGPT understands me very well.

During the process, it offered numerous suggestions including 1,500 small book ideas and referenced authors, titles etc which I saved as well.

While we conversed I was reminded of a sequel I created to War of the Worlds by H.G. Wells. So I'm going to reprice the book and ebook and begin an Amazon Ads Campaign for those.

I have over 1600 titles published and its hard for me to remember what I have out there. I suppose that's a nice challenge to have. My poor brain!

Today also I will take the time to think: Wouldn't it be great if I could do such and such with public domain titles?

Then I will work with Codex and ChatGPT to build a program/system to do just that.

An obvious challenge that I know have is not how to find titles to publish since I now have thousands I can access, but how to speed up the process of publishing.

Another challenge to discuss with ChatGPT about is that KDP Amazon now has daily limits to how many titles we can upload. How to work around that? Well one way, of course, is I can move titles over to Wen and then she can max out what she is allowed.

It will interesting to see what ideas we can come up with.

What I am doing with public domain titles YOU can do with anything else.

Don't hesitate. Just dive right in and get working with ChatGPT and/or Claude which I also use a lot.

My Publishing Empire is truly getting ready to explode and I will not take NO as an answer.

To Our Success!

Mel Waller