Show Notes
Two founders sit down on a Friday with the World Cup playing in the background, and within ten minutes one of them casually reveals he has built a version of himself that works while he sleeps.
That is the hook, and it is not hype. Jason Katz, co-founder of Kindling Solutions, walks through what he calls his personal content machine: a chain of Notion databases, AI agents, and approval triggers that takes a single spoken idea and turns it into finished video, social posts, and carousels, all before he sits down at a computer. The genius is not the automation. Plenty of people automate. The genius is that the output sounds exactly like Jason, because the system is engineered around authenticity instead of around shortcuts.
Here is the part that should make every founder lean in. Jason does not let the AI write his ideas. He lets the AI interview him. He talks into his phone in the backyard with a coffee, an interviewer agent trained on the tactics of Joe Rogan, Oprah Winfrey, and Howard Stern pulls his real takes out of him across ten to twelve questions, and only then does the structuring begin. The words are his. The machine just gives them shape. As he puts it, the context truly does half the work, and that is the line nobody is saying out loud.
Meanwhile Ryan turns the conversation into a masterclass on performance itself. After more than a thousand podcasts, he has reduced great content to a few unglamorous truths: sleep and caffeine are the real production stack, clarity beats cleverness, lead with a current event so your guest can find their feet, and tell yourself to speak ten percent slower so the ums take care of themselves. It is the kind of advice that sounds obvious until you realize almost nobody actually does it.
Both threads land on the same destination. First-time founders obsess over product. Second-time founders obsess over distribution. Jason and Ryan are both, by their own admission, finally crossing that line, moving from "what is this business" to "let the world know what is up." The episode is the sound of two operators getting comfortable being the face of the thing they built.
Named Frameworks
The Personal Content Machine
Jason's end-to-end pipeline that turns a spoken idea into finished content without a desk.
- →Feeder databases pull raw inputs: a news finder, a transcript puller, a free-form idea capture you talk into on your phone, and an expert-takes feed that watches other creators.
- →Status changes act as triggers. Flip an idea to "Selected" on your phone and a watching routine fires the next step automatically.
- →Routines move work down the line: research, then interview, then essay, then scripts, then editing, then review, with each stage stamping a new status when it finishes.
- →Human touchpoints are deliberately rare and mobile: pick an idea, get interviewed, tap a checkmark in Slack to approve. Nothing requires a desk.
The Interview-First Authenticity Loop
How the system captures Jason's real voice instead of inventing one.
- →An interviewer agent trained on great interviewers extracts your genuine takes before any writing happens.
- →The draft is built from your interview answers, not invented from a prompt.
- →A reference doc about your audience keeps the agent on target so it stops probing aimlessly.
- →The result is content that feels like you, because it is you, just structured.
The Three-Lens Research Team
A trio of agents that makes every idea honest and durable before it gets written.
- →Lens one finds proof points that support your idea.
- →Lens two gathers supporting evidence and data.
- →Lens three actively hunts the counter-take, steel-manning the opposing view so the final piece holds up.
The Writer's Council, a 9-Out-of-10 Gate
A quality bar enforced by a panel of agents, not a single pass.
- →A panel of agents, each trained on a different famous writer, grades every essay.
- →The piece must score nine out of ten across all of them to pass.
- →The council polishes structure and argument only. It never replaces Jason's ideas or words.
Ryan's News-First Interview Arc
The shape of a great interview, refined over a thousand episodes.
- →Open with a relevant current event so the audience feels it is timely and the guest gets ten to fifteen minutes to relax and build trust.
- →Move into the business once the trust is there.
- →Weave in the personal and the KPIs, the run rate, the revenue per employee, the nitty gritty.
- →Pay it off near the end so the listener walks away genuinely understanding the person.
The Performance Stack
Ryan's unglamorous truths about being good on camera or on a podcast.
- →Sleep and caffeine are the two biggest performance variables, and balance matters more than volume.
- →Aim for clarity and coherence. Ask the simple question simply.
- →Speak ten percent slower to cut filler and add warmth.
- →Keep your energy slightly above the guest, never competing, always letting them be the star.
- →Chase tangents with curiosity, then bring it back.
Founder Experiment
Pick one piece of content you would normally just sit down and write. Before you write a single word, build a one-page reference doc that answers exactly who your audience is and why they pay you with their money or attention, to be entertained, to learn, or to get a problem solved. Then have an AI interview you for ten to twelve questions about your idea, talking out loud, not typing. Let it draft only from your answers. Compare that output to how you normally sound. The point is not speed. The point is to feel how much the upfront context does the heavy lifting.
Glossary
VoiceTrack
PromptSmart's patented feature that scrolls a teleprompter at your natural speaking pace and pauses when you go off script.
Human in the loop
A workflow where automation runs most steps and a person approves or intervenes only at key checkpoints.
Feeder database
A source table that collects raw inputs, such as news, transcripts, or captured ideas, before processing begins.
Routine
An automated watcher, referenced in the episode as R3 and R5, that triggers the next action when a status changes.
Steel-man
Building the strongest possible version of an opposing argument, the opposite of a straw-man.
B-roll
Supplementary footage cut over the main clip to illustrate what is being said.
Carousel
A multi-slide social post that walks through an idea slide by slide.
Whisper
A speech-to-text model used here to generate captions.
Remotion
A tool for building videos programmatically, used here for editing and captions.
GoHighLevel
An all-in-one marketing and CRM platform used here for publishing.
Claude Code
Anthropic's agentic coding tool, used to build and run the system.
Run rate
The annualized projection of current revenue, as in "annual run rate."
ICP
Ideal customer profile, the specific buyer a business is built to serve.
Questions This Episode Answers
- →How do I produce a high volume of content without it sounding like soulless AI slop?
- →What is the actual stack and sequence to build a content system that runs without me at a desk?
- →How do I get AI to pull my best ideas out of my own head instead of inventing fake ones?
- →I just left my job and feel like a fraud representing my own company. How do I get past that?
- →Now that I can build anything, what should I actually build, and how do I get it in front of people?
- →How do you keep AI agents from going off on tangents?
- →What actually makes someone good on camera or on a podcast?
- →Should a new founder focus on product or distribution?
URLs Mentioned in the Episode
- Kindling Solutionshttps://kindlingsolutions.com
- Jason Katz on LinkedInhttps://www.linkedin.com/in/jasonkatz99/
- PromptSmart (VoiceTrack)https://promptsmart.com
- Remotionhttps://www.remotion.dev
- GoHighLevelhttps://www.gohighlevel.com
- Claude Codehttps://claude.com/claude-code
- Ryan Estes on LinkedInhttps://www.linkedin.com/in/estesryan/
- AI for Foundershttps://aiforfounders.co
