My wife was working at our dining table one morning and asked if I could help point her in the right direction on getting a blog up. They’d been trying for two years, but their platform doesn’t support blogs natively. And for SEO, you can’t just throw it on a separate domain. It needs to live on the company’s main domain so search engines treat it as part of the same site. That means a reverse proxy: the blog actually runs somewhere else, but when a visitor clicks through, it looks and feels like they never left. Getting that right is why it kept stalling.
I had some time that afternoon so I put something together.
Three hours later she was looking at a working proof of concept that matched their branding so closely she asked how I got their colors. I told her I found their public API.
“You should text this to Mimmi,” she said.
Mimmi is the owner, and I sent her a text.
“Hi Mimmi, this is Trey. Annie mentioned you’d been thinking about adding a blog. I had some time after lunch and put something together.”
Then the link.
She figured out the business value on her own and booked a meeting before the day was over. I didn’t have to sell her on anything.
Here’s how the three hours actually went.
Annie showed me a competitor’s site, a company running on the same platform. They had a blog running, which told me someone had already solved this problem. We could see slight inconsistencies between their blog and their main site, small differences in spacing and styling. Not perfect, but proof it was possible.
Then I turned to their site. Found their public API. Pulled their sitemap, color schemes, logos. Now I had a blueprint.
I opened Claude Code and started building. I’ve never built a WordPress site and I sure as hell don’t know how to make one site look identical to another. But that’s exactly what this needed: when someone clicks from the main site to the blog, the experience can’t change. And when they’re on the blog and click to go back to the main site, that has to work too. Same header, same footer, same colors, same navigation. If the visitor notices the transition in either direction, you’ve failed.
I fed in the API data, then took screenshots of the actual site and gave those to Claude Code so it could see how things were organized: header structure, footer layout, the spacing. Moved things around based on what I was seeing.
Within about an hour it looked damn near identical to their site. Branded, functional, sample posts seeded. Not a wireframe. A real working blog.
I put it on a live URL and sent Mimmi the link.
After she replied, I followed up.
“There’s a bit more to do before it’s production-ready. Proper hosting, coordinating on the proxy, QA, that kind of thing. Nothing crazy but worth doing right. I’d love to sit down with you sometime soon and walk through that, plus a few other ideas I have.”
I mentioned I thought there was a real opportunity to look at the business more strategically and find some easy wins. And that the demo was on a temporary server, so it wouldn’t be around forever.
Not a pressure tactic, just a fact. She booked a meeting the next morning. The whole thing happened in one text thread: demo sent, enthusiasm confirmed, broader conversation planted, meeting booked.
The next day we got on a call. Talked through the blog, talked through some other things I’d noticed about the business. We got to the part where someone has to say a number. I quoted a fair price. Done. I sent a scope of work that evening and she replied in three minutes: “Lets do it!!”
That created a new problem: how do you legally charge for work in Sweden? I’d been living here for a few years but had never run a business here. So I figured it out. Applied for an enskild firma, registered for VAT, built out the contracts and invoicing system.
Twenty-four hours after that first text, I had a deal. Within two weeks, I had a registered consulting practice.
I’ve been thinking about that afternoon a lot.
There was no brief, no discovery process, no scope document, no agency onboarding. Annie mentioned a problem and I built a solution. The demo existed before there was a client. When there was a client, the demo had already done most of the selling.
The speed wasn’t unusual for me anymore. That’s the part worth explaining.
A year ago I couldn’t have done this. Not because I didn’t have the ideas, because I’ve always had the ideas. But I didn’t have the execution capacity. I’ve never been a developer and I’ve never written code in my life. What I’ve learned is how to manage AI like a team, and when you know how to do that, the distance between “I have an idea” and “here’s a working demo” collapses fast.
That’s what this site is about. The process, the failures, the stuff that actually compounds.
It started with an afternoon I wasn’t even planning to work.
Get notified when I publish.
What worked, what didn't, what I'd do differently. Delivered to your inbox.