Skip to main content
NEW WEBSITE & SWAG

NEW WEBSITE & SWAG

Posted by Beren McKay on Apr 11th 2026

Journal · Founder Update

Behind the Build

How One Engineer
Rebuilt His Website
in 20 Minutes

With the MERINO DM launching later this summer, I knew the site needed to be ready for it. New product, stronger line-up, more eyes arriving — the existing site wasn't going to carry the weight. So over the last several months I've been working on a full refresh: new content, better visual structure, and a cleaner way to tell the Pepperwool story.

While I was at it, I finally did something about the swag requests. I've been getting asked about t-shirts and stickers for a while now. I partnered with Stickermule, put together a shop on their platform, and uploaded the designs. They handle the printing and fulfillment — I just had to make things worth putting on a shirt. I've been wearing them for a few months and I'm genuinely pleased with how they turned out. You can find them under SWAG on the site.

But the question I've been getting asked more than anything else — more than the DM specs, more than the swag — is about the site itself. How did one guy with no web development background end up with something that looks like this?

Fair question.


The Short Version

I Used AI.
But Not the Way You Might Think.

I'm an engineer by training and by instinct. I understand systems. I understand the logic of how things fit together. What I am emphatically not is a web developer — and I don't have the budget that custom web development typically requires at this level of finish. So when it came time to rebuild the site, I had to find another way.

The answer was AI. But the part people assume was easy was actually the part that took the longest.

  • Six Months

    Learning to tell the story. Before I could ask AI to build anything, I had to know what I was trying to say. I spent six months working through the brand — what Pepperwool stands for, how to talk about the products, what the founding story actually is and why it matters. This wasn't AI work. This was thinking work. The AI needed something to work with.

  • One Month

    Learning to tell a story on a website. A good story in prose and a good story in HTML are different things. I spent a month understanding structure — how pages flow, what sections do what job, how a visitor moves through a site and what they need at each moment. I didn't write any code. I just learned what I needed to ask for.

  • Several Weeks

    Building iteratively with AI. I described what I wanted. AI output HTML test pages. I reviewed them, made notes, pushed back, refined. This happened dozens of times across every section of the site. Each iteration got closer to what I had in my head. The AI was doing the coding. I was doing the directing.

  • One Upload

    The final assembly. I gathered everything — colors, fonts, copy, stories, all the page mockups from the iterative work — and uploaded it into AI with one instruction: build me a complete website theme based on all of this. It processed seven months of work and output a coherent, finished theme.

20 min

That's how long it took to generate the complete custom theme — once seven months of preparation were in place. The 20 minutes is real. So is everything that made it possible.

Then I spent two days on cleanup. Padding around images. How things looked on mobile versus desktop. Small misalignments that only showed up in the real environment. The kind of detail work that no automated process gets perfectly right on the first pass.

And then it was done.


What I Actually Think About This

It's Wild.
And It Isn't Magic.

I'll be honest: the fact that I was able to produce this site is genuinely remarkable to me. Six months ago, a custom site at this level of fit and finish would have required a developer, a designer, a budget I don't have, and a timeline that doesn't fit how I work. None of those things are true anymore.

But the thing that made it work wasn't the AI. The AI is a capable tool. What made it work was having something real to put into it — a story I understood, a visual identity I could describe, a structure I had thought through carefully enough to direct. AI didn't know what Pepperwool is. I had to tell it. And telling it well required everything that came before the 20 minutes.

The honest summary

If you try to use AI to build a website without knowing what you're trying to say, you'll get a website that says nothing. The preparation is the work. The AI is what makes the execution achievable for one person with no developer budget and no developer skills.

I'm thinking about doing a longer write-up on the full process — the specific tools, the prompting approach, the things that didn't work before the things that did. There are a lot of small businesses and solo founders out there who are in exactly the position I was in six months ago, and I think a detailed account of how this actually went would be genuinely useful. If that's something you'd want to read, let me know.

For now — the site is new, the DM is coming, and the swag is real. I'm pleased with all of it.

Beren

Common Questions

If You're Wondering
How This Works

What AI did you use to build the website?

Claude, made by Anthropic. I used it throughout — for working through the brand story, for generating and refining HTML page mockups, and for the final theme assembly. The process was conversational: I described what I wanted, reviewed what came back, pushed back on what wasn't right, and refined from there. It took many iterations across many sessions to get to a point where I could hand it everything and ask for the final output.

How much did it cost to build the site with AI?

A fraction of what a custom development project would have cost. The AI subscription is a few dollars a month. The real investment was time — specifically the months I spent before touching the AI, learning what I needed to say and how to direct the output. If you go in without that preparation, you'll spend that time inside the AI instead, and the results will show it.

Do you need coding skills to build a website with AI?

I don't have them — not in any meaningful sense. I can read HTML well enough to spot problems, which helped during the two days of cleanup after the theme was generated. But I didn't write the code. What I needed instead was clarity about what I wanted: the story, the visual identity, the structure of each page, the specific decisions behind each design choice. The AI handles the syntax. You have to supply the thinking.

Could anyone do this, or does it help to be an engineer?

Honestly, the engineering background helped less than you'd think. Systems thinking is useful — breaking a website into components, understanding how the pieces connect — but it's not required. What matters more is being able to describe what you want precisely enough that the AI can act on it. That's a communication skill more than a technical one. The people I think would struggle are the ones who haven't done the work of figuring out what they're trying to say before they start asking for it.

Will you write a more detailed breakdown of the process?

I'm planning to. The short version is here. There's a longer version — with the specific approach to prompting, the things that didn't work, the order of operations that mattered — that I think would be genuinely useful for small business owners and solo founders who are trying to figure out whether this is real or hype. It's real. It's also not as simple as people imply when they say "just use AI." I'll write it properly when I have the time to do it right.


Now Available

Pepperwool swag — t-shirts, stickers, and more — is live on Stickermule. Find it under SWAG in the site navigation. I've been wearing the shirts. They're worth it.

Coming Summer 2026

The MERINO DM

2.9" blade. Full hand on the handle. The same carry philosophy as the MM in a daily-sized frame.

Learn More

Get the latest updates on new products and upcoming sales