I built a machine that turns one interview into a blog post, a book chapter, and a LinkedIn launch.
How I built a content system that turns one Riverside interview into ten-plus published assets — a GEO-optimized article, a LinkedIn launch post, a book chapter, and more. The Pre-purposed machine, explained.
Key Takeaways
- Recording was the cheapest line in the budget. Plan, edit, and format were eating ~90% of the time on every interview cycle — so the machine attacks every step that isn't recording.
- One Riverside interview now produces 10+ assets: a long-form video, a GEO-optimized article, a LinkedIn launch post, and a book chapter. Same recording, no new shoots, no new writing rounds.
- The orchestrator runs four phases — Article, LinkedIn, Shorts, Publish — with a human review gate between each. Nothing ships blind.
- A book serves as the structural spine. Every article carries a chapter slug; the /book page lights up chapters automatically as posts land.
- Across 8 cycles, output rose +800% per interview with no extra recording days — same input, more output, quality held.
The hardest part of B2B content isn't making the thing. It's making the thing once and getting twelve uses out of it. So I stopped treating each interview as a deliverable. I started treating it as raw material for a system that publishes itself.
I built this because I needed it. Most teams need it too.
I work with B2B marketers all day. Most of them sit somewhere on the same ladder. Novices wing every shoot. Operators ship on a calendar but can't tell you why. Strategic teams pre-purpose. The teams at the top treat video as the default unit of communication.
Where teams sit on the video maturity ladder · the machine ships at the right end
This essay is about the machine I built to sit at the right end without hiring an army to do it.
The old way was a tax I kept paying.
Most teams ship a video, write a blog post about it three weeks later, post a clip to LinkedIn whenever someone remembers, and call that "repurposing." It isn't. It's redoing the work, in a different format, with a different person, on a different deadline. Every step is a fresh decision. Every step has a tax.
Here's what the round-trip used to cost me, broken down across a single interview cycle. Most of the time wasn't in the recording. It was everywhere else.
Where time used to go on one interview · recording (orange) was the smallest part · everything else was the tax
The interview itself is the cheapest line in the budget. Everything around it was killing me. So the machine focuses on every step that isn't recording.
One input. That's the rule.
The machine eats exactly one thing. A Riverside interview. The recording is the only manual work I sign up for. Everything downstream is scripted, reviewed, and shipped by Claude Code.
What comes out of one interview · the orange one is the only count that matters
One interview produces, today, a long-form video, a GEO-optimized article, a LinkedIn launch post, and a book chapter. By end of year, also: shorts, landing pages, sales clips, infographics. Same recording. No new shoots. No new writing rounds.
The orchestrator runs four phases.
I call the front door interview-to-everything. One command. It takes a transcript file, a guest name, a YouTube URL, and an optional book chapter slug. It walks through four phases and pauses between each so I can review before anything ships.
Phase A · Article
The transcript becomes a 1,200-word article with H1, H2s, an FAQ block, and JSON-LD schema. It's tagged with a chapter slug from my book. It's structured to get cited by ChatGPT and Perplexity, not just ranked by Google.
Phase B · LinkedIn
The published article becomes a LinkedIn post. Hook, body, first comment, image brief. The post launches the interview. The article is what people click through to read.
Phase C · Shorts
Every short clip I cut gets its own landing page, optimized to answer one question an AI engine might ask. Not built yet. Queued for V2.
Phase D · Publish
The article goes to disk as markdown, opens a branch, opens a pull request, and hands me a Netlify preview URL. I click around. I merge. The site updates. I never touch a CMS.
Most teams ship a small fraction of what they record.
Here's the painful truth. Out of every hundred minutes of B2B video most teams record, only a tiny fraction ever lands in front of a buyer in a useful, findable form. The rest sits on a Vimeo somewhere.
14 of every 100 minutes recorded ever ship as a published, findable asset · the rest is waste
The machine raises that number. Not by making people record more. By making everything you already recorded actually get published.
The book is the spine that turns repurposing into compounding.
Here's the part most people miss. A repurposer just spreads one thing across channels. A machine compounds when every output has a place to live that's bigger than itself.
So I gave the system a spine. A book. Pre-purposed: The B2B Leader's Guide to Video Content That Compounds. Fourteen chapters across five parts. Every article I publish carries a chapter slug in its frontmatter. The book page reads that and auto-lights-up chapters as posts land.
The book writes itself one interview at a time · 4 of 14 chapters in motion after one weekend
I'm not writing a book on the side. I'm writing it in the act of doing my real work. Every interview I'd do anyway becomes a brick. The book writes itself by me doing my job.
Output goes up. Time per asset goes the other way.
I track this. Every interview cycle, I log the total time from recording to publish, and I count the assets that came out the other end. Here's the trend over the last few cycles, while the machine has been assembling itself piece by piece.
Assets shipped per interview, cycle over cycle · +800% across 8 cycles, with no extra recording days
That's the line that matters. Same input. More output. Quality held. That's not a productivity hack. That's a structural change in how content gets made.
Nothing publishes blind.
The thing that kills most "AI-powered content systems" is the same thing that killed marketing automation in 2015. Volume without quality. Stuff gets published nobody read first. The brand starts smelling like a content farm.
So the rule on this machine is non-negotiable. Every run produces a branch, a preview URL, and a pull request. I see it before the world does. If it's off, I fix it or I kill it. Then I merge.
What's coming next.
V1 of the orchestrator is queued. Four to six hours of focused build. After that, the shorts pipeline, then thumbnail generation, then infographic auto-drafting from the article itself using the same primitive library that built the graphics in this very post.
What to do this week, if you're in this fight.
Pick one piece of content you ship regularly. Interview, podcast, post, demo. Write down every step between recording and published. Count them. Most teams find twelve to twenty. Then ask which steps a machine could do tonight if you let it. That list is your first sprint.
Want me to map your system the same way? We do this with B2B teams who already have video and want it to compound instead of accumulate. 30 minutes. You'll leave with a wiring diagram of what you have, what's leaking, and what to build next. No pitch. If we're not the right partner, we'll say so.