by /
I sat on a library of 260+ YouTube videos for a year and a half after Dennis Yu gave me the same directive we now give every Local Service Spotlight client: claim your Google Knowledge Panel, then repurpose everything you have already published. The panel got claimed. The repurposing stalled, because by hand it cost an hour per article. This is the Claude pipeline that finally cleared the backlog, documented so you can run the same play on your own library.
The pattern is the same one we used when we inventoried Paul Ryazanov’s content with an AI agent and later automated his content pipeline end to end. All of it serves the principle Dennis teaches in Own Your Name on Google: your name should resolve to one strong entity, supported by content you control.
The library, counted
The inventory came first because scale is what killed the manual approach. My personal channel has 1.9K videos. The Dunk Talk Podcast has another 141, with 69 full episodes. My channel Digital Marketing w/ Dylan Haugen has 44. And I’ve been a guest on 8 other podcasts. Claude went through all of it and we marked what was worth repurposing: 143 personal videos, all 69 episodes, all 44 marketing videos, and the 8 guest spots. That’s 264 articles from one spreadsheet.
One tracker sheet, four tabs
Everything lives in a Google Sheet with a tab per corpus: Personal Videos, Dunk Talk Episodes, the marketing channel, and Guest Appearances. Each row carries the video title, the YouTube link, the article link once it exists, and a QA’d flag. The podcast tab adds episode numbers. That one sheet works as the assignment queue, the status board, and the QA log at the same time.
Training the voice before scaling the volume
The first two episode articles did not ship on autopilot. I went through them section by section and told Claude exactly what I would have said differently. Those corrections became skills, a persistent knowledge base describing how my articles should read. That step is the one most people skip, and it’s the difference between articles that sound like a model and articles that sound like the person.
“It’s actually taking the transcript itself from the video, and then it is taking that transcript and repurposing it into an article, taking out a bunch of key important details from throughout the video. It’s even quoting people.”
That’s from the video above, and it’s the standard every article in the queue is held to. The source is the transcript, and the transcript is the boundary.
Publishing without touching WordPress
Publishing runs through Claude Code with a WordPress application password. Claude writes the article, embeds the source video at the top, sets the slug, the meta description, and the Rank Math optimization, then edits the post directly on the site. The next piece is Google Sheets API access so the tracker updates itself as each article goes live. The old workflow, ChatGPT plus custom GPTs plus manual WordPress work, cost an hour per article and eventually got abandoned. This one costs minutes of human attention, and most of those minutes are QA.
The E-E-A-T line we won’t cross
Google added the second E in E-E-A-T, experience, specifically because AI made mass generated content cheap.
“They added the second E, experience, because they knew that stuff like this would happen where people would just start mass generating content because of AI tools. That is not what we do.”
Repurposing is the opposite of generating. Every article traces to a transcript of something that actually happened, with the real names, real numbers, and direct quotes kept intact. That’s what makes the content corroborate the entity instead of diluting it. When I interviewed Jordan Kilganon on episode 47 of Dunk Talk, that relationship became public, linkable evidence of who I am and who I work with.
What it feeds: the Knowledge Panel
My confidence score in Google’s Knowledge Graph is 208 as of this video. Every published article adds corroborating information about the same entity on a domain I control, and the score has climbed steadily as the articles have gone out. The entity side has to hold up its half too: when my panel dropped out of search results this year, a Wikidata repair restored it, and I wrote a first-person account of that scare on my own site. You can check any name’s entity and score with the Knowledge Graph Explorer.
The build pattern
Inventory, voice, publish, track. The skills get sharper with every QA pass, and each article strengthens the entity the next one points at. If you have a video library gathering dust, start with the inventory sheet; Claude can build it in an afternoon. My own walkthrough of the system covers the personal side of the same build, and the live results are accumulating on dunktalks.com right now.
Dylan Haugen
Dylan Haugen is a professional dunker, content creator, and editor at the Content Factory, where he transforms podcasts and interviews into strategic brand assets. He collaborates with Dennis Yu to support young entrepreneurs and business owners in building their personal brands through education, transparency, and effective content marketing. As the host of the Dunk Talk podcast and a dedicated advocate for establishing dunking as a recognized sport, Dylan combines athletic expertise, storytelling, and digital strategy to help elevate the next generation of creators.