There’s a lot of discussion about what AI might do. I’m generally more interested in what I can make it do on Tuesday.
Here are five things I’ve actually used AI for recently inside my business.
1. Turning years of email content into a blog
I had somewhere around 20 essays that I’d previously written and sent to the email list, but copying and pasting that essay onto the site was a bridge too far for me.
Normally that would mean somebody going through every essay, creating a WordPress post, reformatting it, adding the image, setting the publication date, filling out the SEO information, and publishing it.
Once I hit “Go” on emails, I want to be done with them. The fun, creative work is over.
Instead, I connected ChatGPT to our WordPress site and gave it the essays. To do this, I enabled the WordPress plugin in the ChatGPT app, then I logged into WordPress and enabled MCP to update, add, delete, and change posts and pages.
It reformatted them for the site, handled the WordPress structure, added the relevant Yoast SEO information—including focus keyphrases, metadata, and search descriptions—uploaded the associated images, and published each article using the date on which the original email had been sent.
This backlog, which would easily have eaten up half a day or a full day of mindless work (which is why I didn’t do it), was handled in one go. If you’re curious how I did quality control, the WordPress plugin forced me to manually approve the changes by messaging “Approved.” So I did have to babysit the computer, but I babysat it while doing other stuff.
2. Having AI audit and actually improve our SEO
I know enough about SEO to know that it’s important, but I do not know enough about SEO to want to manually audit an entire website. Nor do I want to become an SEO expert. The other problem is that it’s extremely difficult to know who’s actually good at SEO and who’s faking the results. So, like publishing the essays to the blog, my solution for half a year was to avoid it.
I gave ChatGPT Codex exports from Google Search Console, Google Analytics, and Microsoft Clarity, then I had it crawl the entire website as if it were a search or AI bot.
From that, it created roughly 50–60 recommendations covering both the website as a whole and individual pages.
But then my new problem was that I didn’t want to do the work to make the changes either.
Using the same WordPress connection, I had the AI go in and start implementing the fixes.
It updated metadata. It improved page structure. It adjusted headings. It found and cleaned up dead links. It improved focus keyphrases and descriptions.
It also created a custom llms.txt file designed to make the structure of our website easier for AI systems to understand.
All these SEO changes aside, I also brought in a human to improve our home page and speed up the site. I, personally, probably spent a total of 20 hours working on SEO over the course of 5 weeks.
Our SEO score measured by Ahrefs improved from 10 to 67 (out of 100) in 5 weeks. Not bad.
3. Using our website data to decide what to redesign
We also took our analytics data and tried to answer a different question:
What are people actually doing when they come to the site?
We mapped common visitor paths and looked at where people were spending time, where they were clicking, where they were leaving, and how they were reaching different parts of the website.
That analysis helped inform the broader home page redesign I mentioned above.
We brought back our designer. Step 1 was to make the site way more fun and funky. We never, ever want someone asking us, “What AI built your site?”
We completely redesigned the homepage, added tons of animations, and made significant changes to a dozen other pages based on what the data suggested people were actually trying to do.
Within the first week, we were already seeing meaningful improvement in things like traffic, time on page, and clicks.
In this case, AI was the site analyst who helped us understand where the site was failing and where human design and development effort would have the biggest payoff.
4. Giving new hires safer AWS permissions
AWS is an absolute disaster to understand if you don’t work in it every day.
When you give someone access to AWS, you may be giving them access to some of the most critical infrastructure and data in the company.
Applications. Databases. Data lakes. Domains. Production infrastructure.
The lazy solution is to give someone broad administrative access because figuring out exactly which permissions they need is painful. This is what most SMBs do.
It is also a terrible security practice.
I recently needed to provision access for a few people, and I used AI to help create a least-privilege permissions structure. In other words, AI helped me give each person enough access to do the job, but nothing they don’t actually need.
Doing that manually would’ve been a nightmare.
With AI, I was able to work through it in a couple of hours instead of defaulting to over-permissioning or spending days researching IAM policies.
5. Automating our weekly business-review agenda
Every week, Bill, my director of ops, and I have a business-review meeting.
Previously, the agenda was mostly whatever happened to be top of mind.
I used AI to create a consistent agenda that works its way through the important parts of the business. Then, I automated the administrative work around it.
Each Monday, ChatGPT scheduled tasks takes the prior week’s agenda and creates the new one.
It moves the previous updates and action items into a historical section. It preserves who owned each item and when it was supposed to be completed. Then it creates clean spaces for this week’s updates and this week’s new action items.
Now we can very quickly see:
What did we say we were going to do? Did we do it? What’s changed? What needs to happen next?
I understand that this is a pretty elementary automation and that I really should’ve had an agenda for the last 4 years, but it was one of those open items that I never resolved.
It probably saves 15 or 20 minutes a week. That’s enough of a win for me.
There’s no pattern that I’m paying attention to or “summing it up” to this. AI SEO is something many people are trying to figure out, and I’m happy to share more with you about that if you want (just reply letting me know you want it). I’m not an expert in any of these things, but I am an AI power user, and so I figured I’d share how I’m using it lately.