I went on The Edward Show to walk through how I build linkable assets with Claude Code, and the first thing Edward asked about was my programming experience. My answer was one word: “None.”
The episode is number 1,224, which means Edward has put out a show every day for 1,224 days in a row. He asked me on after I told him I was the one who built SEO TV, a searchable video store of SEO YouTube channels and podcasts. The whole episode is on his YouTube channel.
The second take you see on YouTube
The second take you see on YouTube is there because Edward stopped the first one halfway through. I had such a fun time on The Edward Show, but it started rough. I have watched it every day for over a year, and I credit it for the SEO I know that actually moves the needle, so being asked on was a really big deal for me.
It was also my first podcast since I started doing SEO two years ago, and I had not been on one in a long time. I was rusty. My presentation was full of SEO and the back and forth between David Quaid and me, and Edward wanted a complete beginner’s guide to vibe coding a website that doesn’t look like AI made it.
Edward is such a nice guy that he gave me a few hours to set it up again. Public speaking has always been a bit daunting for me, but like I told him, I wasn’t going to miss this for anything.
Why linkable assets are worth building
Linkable assets are worth building because people share them and link to them, and the one main link on the page points back to your business. Edward put it better than I did: when something is in the news, people spin up a site that rides the topic, make it funny or useful, and the attention flows to whoever built it. You can launch one on Product Hunt or send it to journalists.
How SEO TV started with one line
SEO TV started with one line typed into Claude Code: “let’s plan a topical map for a new domain seo-tv.com.” That was the whole prompt, because by now Claude knows what I mean by a topical map. The idea came from David Quaid, who owns the domain and picks the channels, so someone new to SEO lands on good advice from the start.
The map took about half an hour. Claude cast a wide net on SEO video searches and came back with this:
| Step | Count | What it means |
|---|---|---|
| Related searches found | 21,180 | everything Claude pulled from the seed terms |
| Worth targeting | 213 | after cutting the off-topic and the overlaps |
| Pages planned at launch | 73 | one topic per page, so none compete with each other |
| Pages as of the episode | 139 | added as earlier pages got indexed |
Before each page got built, Claude checked what Google shows for that search. Google doesn’t serve videos on every query, so the pages that didn’t get video results didn’t get a video block. The whole build took about five hours, and a lot of that time was me waiting.
What makes SEO TV more than a static page
What makes SEO TV more than a static page is a small script on the server that checks YouTube every hour and adds new releases on its own.
The search box runs in your browser with no API, so it’s fast. The pages are built as plain HTML on my PC before they go up, so Google and a Screaming Frog crawl see every page and every title right away.
How I set up Claude Code for people starting out
How I set up Claude Code for people starting out is simple, and none of it is technical. Get the desktop app and use the Code tab, not the chat on the website. The free plan doesn’t include Claude Code, so start with the $20 plan and see if you like it.
Then:
- Make a project folder for each site, so Claude stays organized.
- Tell Claude to keep a notes file in every folder and read it first, so it knows where you left off.
- Start on Ask permissions, so you see every button it wants to press. Move up once you trust it.
- Ask for three or four designs before it builds anything, and say “this sucks, give me something else” when it does.
No matter how sure the AI sounds, you make the final call. The exact prompts I use to get Claude to build sites like these are on my prompts page, so you can start where I am now instead of where I started.
Avoiding the AI website look
Avoiding the AI website look was the mission that got me here. If you have used AI for a while, you can spot its sites from across the room: the same clean template, the same fonts, fake photos, pills, rounded cards, little eyebrow labels over every heading, and “it’s not this, it’s that” slogans.
Tell it no, and keep telling it no until you like it.
What works for me is reverse engineering the top sites in a niche, copying what’s good and making it better. Real photos and screenshots beat generated images. One big idea per page. I like big type and artsy layouts. I also use three free design skills from GitHub: Hallmark, which catches the AI look before it happens; Impeccable, which adds the finishing touches; and UI UX Pro Max, which gives Claude a deep stock of layout ideas.
Every site also gets its own logo, favicon, Open Graph image and share card.
Turning mistakes into rules and skills
Turning mistakes into rules and skills is what made the last few months faster. Every time Claude gets something wrong, it goes in the notes. When the notes on one subject get long, I have Claude turn them into a skill. On the day we recorded I had 811 rules and 59 skills. A lot of them came from pasting Edward’s transcripts into Claude and saying “make a skill from this.”
My most expensive lesson is now a rule. I set up a GTA 6 map site, went to bed, and woke up to Claude still working, 8 hours and 46 minutes in. What it made was cool, but it put the game’s places on a map of Florida, which is not what I wanted. If you’re not sure, Claude is not sure, so say exactly what you want.
The opposite surprise was the 3D vape site, where Claude rendered product images on my graphics card without asking. It downloaded software to do it, which was a little concerning, and I said so on the show.
What cheese.help showed without link building
What cheese.help showed without link building is that a niche reference site can earn links on its own. I bought the domain this summer for $4 because it was a fun name, put up a rough AI site to warm it up, and forgot about it. Weeks later Search Console showed clicks and impressions, so I had Claude rebuild it into a real cheese encyclopedia.
I never built a link to it. When I checked Ahrefs before the show, it reported 11 links from MSN pointing to seven pages, a few food blogs, and about 4,200 AI citations, up from about 3,000 a few days earlier.
Where it stands after the episode
Where it stands after the episode: since September I have worked on 114 sites and launched 56, and each one comes out better because the rules carry over.
Thank you again, Edward, for the second take.
Sources
- How to Build Viral Linkable Assets with Claude Code (Edward Sturm, YouTube)
- The Edward Show (Edward Sturm)
- SEO TV
- cheese.help
- My prompts for building websites with Claude (aka.bike)