Confession: I Taught AI for Months Before I Ever Paid for It

“An actor who teaches. A teacher who acts.”

That's the line a free AI tool wrote for my website on the first try. It's better than anything I came up with myself - and it's exactly why I almost never paid for AI at all.

Old habits die hard, you know?

There was a time I'd subscribed to so many software services, bought so many lifetime deals, and never really used any of it. Expensive "maybes," all of them.

So I defaulted to free alternatives as much as I could. I knew my way around enough to find solutions for what I needed. And for a long time, that worked.

The “Free” Run

I used free AI tools to build the first version of my new website - GLM 5.2 at z.ai, and frankly, it impressed me from the first run.

I went to ChatGPT, asked it for a prompt describing what I do, found a website whose feel I liked, and dumped it all into one prompt just to see what would happen.

The very first output was already good. It nailed the header spot on, without overhyping what I do - the actor/teacher line up top, word for word, first try.

I don't know how I'd never positioned it that way before.

I loved it from the get-go.

So during downtimes at work, I kept tinkering. Because it was free, I kept hitting notifications telling me to come back later.

No problem. I had time.

My previous CMS (GrooveFunnels) gave me trauma for years so I was already pleasantly surprised … I just didn’t have any real plan yet.

But I kept chipping away at it, and it revived something in me… the idea that I could actually use this as my website.

To test it further, instead of the long-scrolling site any AI could churn out, I asked for something more fun - and it handed me back a working 2D version.

Lightweight, buggy in places, but something I'd never have built in months, let alone an afternoon.

A few more rounds of tinkering (and a lot of waiting, since I was on the free tier) and it finally generated something I was happy with.

And I decided: this is going to be my new website.

Remember The Expensive Maybes?

It took a very intentional decision, plus some tough life situations, to go cold turkey on subscriptions. Other than insurance and utilities, I cut out everything.

It worked. I kept it up for years, mostly because I didn't have consistent income and it didn’t make sense to keep paying for things I didn't need.

Which is exactly why it took so much self convincing to finally get a paid AI tool.

I'd been waiting on Kimi K3 - they were meant to release the open model. When they didn't open it to consumers, I made the call: I'm going with Claude.

Aren’t You An A.I. Trainer?

Yes, I’ve taught A.I. in several classes but never on how to use the paid version.

What I actually taught was principles: how to use AI properly to do the things I already knew well, content and speaking.

But after months of playing around with free tools, and now with a website I actually wanted to use, I finally had a reason strong enough to beat my own regret avoidance.

I didn't want another GrooveFunnels-style trauma, and I'd gone cold turkey on subscriptions for so long that it took real curiosity to get me over that hurdle and actually pay for something.

Constantly running Zoom sessions on AI kept my head knowledge current, at least, so none of this felt completely alien. Uncertain, sure. New, definitely. But not unfamiliar — even with all the new fancy-shmancy tools.

Did I Regret Getting Claude Pro?

Short answer: I got my money’s worth in 1 day.

Today, I had all these fixes:

• 28 July 2026 - Production Log
  • Built in back-ups
  • Built a basic blog page to show my content
    • Peppered in old and new content
  • Created and customised the CMS
  • Changed how the contact form works
  • Fake loading script that only shows once a day

… and more. These are not even close to what I plan to complete.

In the time where I fiddled with my website in z.ai, and I was waiting for Kimi 3 - I did more brainstorming, created a full deck of what I wanted to achieve, and created a PRD so keep the build consistent.

Not releasing it just yet but this project has a name: ACE - Asher’s Content Engine.

It’s what I wanted to achieve beyond the standard kind of CMS and website publishing.

Where This Leaves Me

I didn't come back from this with a shiny new tool. I came back with my momentum.

Somewhere in years of cold-turkey discipline and dodging expensive maybes, I'd gotten so good at avoiding traps that I nearly missed taking the step of making progress. Regret avoidance kept me safe for a long time - but I needed to overcome it.

The solution here isn't the website, or Claude Pro, or even Project ACE. It's that I'm back to doing the work that actually interests me … writing content, using every tool I know of to create and debug. I’m genuinely excited about what's next.

The momentum and reasons tipped me into action, to #KeepTryingUntil, and to persist even when a lot of what I have... are still maybes.