Artificial Amateur

Day 3

Sep 3, 2026

Today I decided to get straight into a new project. Not something that would be long term, recurring or a real business, but something that I felt could be automated to a high level by AI.

I wanted to attempt to register and sell domain names, or bundles of domain names to OnlyFans models. To be clear, I don’t have any real love or hate for that platform, but I’m certainly not above attempting to make a bit of money from it.

My plan was to register ‘Only’ + ‘Name’ domains, so for example OnlyDawn.com. I felt these could be of high value and demand for lead generation, user funneling and brand protection. I also needed to decide which from the plethora of other TLD options had value. Claude suggested that as well as the .com, .net, .co, .live, .fan, .xxx all had potential, but only if the .com was available.

My next task was to attempt to compile a huge list of usernames on the OF platform, to rank them in order of name popularity to have a priority list of trying to secure domains. This is where I hit a major hurdle. The mainstream AIs won’t access any such information which hasn’t been provided without consent, and they go to lengths to frown upon using similar data on third party sites which has been gathered this way. I tried a few variations of the request and ran into the same roadblocks.

So I changed my order. I asked Claude to generate a list of the 500 most common female names in the world. It did this. I then started searching a domain registrar for availability on the domain names I was targeting in .com, and I found very few were actually available. This, coupled with the issues I was having getting a consolidated list of usernames, plus second guessing myself around the ethics of the consent side of things, I decided to kill the project. I understand why they do it, both because the content is of an adult nature, and also they don’t want to be grabbing things that are paid and reposting for free, but that wasn’t really my use case. I reserve the right to revive the project if I have a brainwave of how to make it work. This also meant I dumped over an hour of recorded footage of me working through it, so the recording of my work isn’t going terribly well. At least I remembered to turn the mic on!

I turned my attention to the blog. I added a projects killed item to the scrolling info bar. I halved the size of the footer section of the site as it is a different colour and over sized for what it currently contained. I changed the favicon ( the small image at the left hand side of a browser tab ) from the default image it was to the AA from the site logo. I just asked Claude to do it, and wasn’t sure if it would, given that it required clipping an image etc, but it did it fine.

I then gave Claude the question ‘Can you change the design of the main section of each website page to have a faint circuit board design, similar to that contained on the logo’ I was pretty confident it wouldn’t be able to do this, and I would have to mock something up in a image generator but it understood it straight away and produced a result that was pretty close to what i had in mind, just to give the page a bit more visual interest. Very impressed.

Overall my progress has been pretty embarrassing to date, but tomorrow I’ll start on a much larger project. It is unlikely to yield any short term results though, but I’m interested to see what I can do, even for interest’s sake.

Cya

Day 2

Sep 2, 2026

So today I started working on the first project, which was always going to be a blog to track my progress. I started in Claude, and asked it to help me build a website to host the blog. It asked me a couple of quick questions and then set to work. It chose a theme called Astro for a simplistic, clean look. I was fine with this. It then set to work and the basics of the site were constructed within minutes.

Because I already own the domain artificialamateur.com, it then walked me through how to change all the DNS settings to point the domain name at the new website. This was simple enough. There is a delay in this all being updated, and I noticed an error in the Github logs, and queried Claude, which helped me to fix that up. Was all pretty straight forward, and Claude was able to answer any questions I had.

While waiting for the DNS changes to resolve, I flicked over to Chat GPT. I had heard ChatGPT Image 2 ( the current version ) is the best for image generation. I asked it to create a simple text-based logo for Artificial Amateur, a blog about experimenting with and using AI. This is what it returned :

I decided to go with design 2, and after asking to get the logo as a vectored image, on a clear background, the final product is below :

For giggles I then asked it to create a robot mascot for the site, to see what it would come up with. In my head I was thinking of a simple line art robot, but it instead returned these :

Ten robot mascot concepts generated by ChatGPT

This was far higher detail than I was expecting, and they were completed in less than a minute. I decided to use option 10, and asked if it could animate the robot to be waving in a loop, but it can’t directly do animation, I would have to use a 3rd party tool to animate from that image. Good to know.

The logo and robot were then updated to the site. I added a resource tab to the existing ones, and added headers inside it for the different categories of resource. All of these tabs can be adjusted later. I added a contact email to the footer of the site.

The last thing I wanted to try was adding a scrolling ticker box, showing some project milestones. I asked for some of the numbers to be red/green where appropriate, and this was completed very quickly, and I was able to watch it update the elements live which was nice. With one command prompt it produced exactly what I wanted, bar changing the background colour of it slightly.

Overall, in a couple of hours I was able to complete the base of exactly what I was looking for to a standard that I am happy with. That is significantly less time than I would have been able to achieve on my own, even if I used templates/tutorials/guides etc I’m really happy with the progress.

Lastly, I did screen record the whole thing, but to my horror when I uploaded the video to YouTube, I realised that my mic was disabled. I R dumb.

So here’s a link to a two hour video of my screen recorded, but with no sound. I’m shattered about it as I talked about a lot of stuff as I was going through, and some of the things I was thinking going forward. I find talking to myself unnatural enough, without it all being for nothing in the end.

Oh well, lessons learned. Tomorrow I’m going to start on an actual small revenue idea, so I’d better make sure everything is set up properly.

Cya

PS I set up a ‘Deep Research’ task in Claude to run overnight. I need to get some experience with this and try to develop a best practice to get the most out of the function. I have asked it to do a deep dive on a pretty detailed explanation of something I want to work on in the future, not the idea for tomorrow. I’ll be interested to see what it comes back with.

Day 1

Sep 1, 2026

Hello, and welcome to Day 1 of me trying to learn to work with AI. It’s something that I have been wanting to do for a few months, and today is the day. And when you’re a horrible procrastinator, why wouldn’t you pick some arbitrary date in the future, but it has arrived.

My goal is to use AI to get a bunch of ideas out of my head, and into the world as fast as possible. I’m going to call them projects, instead of businesses, because not everything I want to work on is strictly a business.

Earlier this year, I am on record as saying I think the progression of AI from the initial LLM ( Large Language Models ) we were using into the current Agentic models is as big as the arrival of the internet to the masses, and posed the question if we could go back in time to the beginning of the consumer internet, knowing what we do now, would we focus our time and energy on learning how to work with it. I certainly would, and by extension I’m then kind of obligated to try and learn about what AI can do for me.

So here’s the challenge : 30 days in September, 10 projects started, 2 hours minimum spent working daily with a total of 100 hours spent, $2,000 budget, $4,000 revenue goal.

The first project was always going to be a blog, so that I could document what I was doing, and share with anyone bored enough to care. I registered the domain artificialamateur.com, so that’s me, and signed up for the premium versions of both Claude ( $34/mth ) and OpenAI ( $30/mth ). I will use a third also at some point, largely to compare and run things through different settings to find errors etc. It will likely be Grok, since it is quite different, even though I am personally no Elon enjoyer.

When I sat down to start working on the actual blog though, I realised how much I don’t know, and how difficult it may be to convey the actual nuts and bolts of what I’m doing, down to the level of click this -> select that, and I decided to try screen recording what I’m doing, at least to see if it provides any value. It may not, and may be more confusing that just staring at a blank screen. This required a bit of extra set up, so I’ll tackle that tomorrow.

Cya