🎙️ Kidnapped Daughters, Raw Audio, and a Deadline

"Jim, if I kidnap your daughters. 

..and tell you that you can't have them back..

until you took a raw audio recording from my phone and matched the quality & editing of whatever I have for my podcast right now, 100% with AI.. how long would that take you?"

That's what Chris Harder said to me.

While I sat across from him for coffee a little over a month ago.

Chris is one of my closest mentors and dearest friends. 

He has a way of cutting through everything and getting to the real question. And the real question that morning was: can you actually build this crazy idea we're having?

I looked at him.

"Give me until tomorrow. And feed my girls well."

We laughed, and before I left, he texted me a raw audio file, I got his intro/outro music, and later that evening I sat down and got to work.

By about 9p, I had something. I texted him a screenshot.

I gave my AI a final prompt with all the context and autonomy it needed to run the the playbook I just made, set a /goal and went to bed. 

By morning, I had:

• 3 versions of his show, all with various hooks auto extracted
• Intro and outro music spliced
• Sound decently normalized
• Light edits attended to
• 5 titles & draft show notes

All waiting for me.

The music spliced in at the right points. Those three hook variations cut from his own voice. 

All from raw audio. All automatic. All done while I slept.

It needed some tuning but DAMN was it awesome. Sent Chris the variations, the assets etc, he loved them.  

We called it what it was right then and there: this is a system that will change everything for podcasters.

And that is how Casthouse 🔗 was born.

🎙️ The Problem I Was Actually Solving

Here is the hard, messy truth about podcasting that nobody talks about. Most shows don't end with a grand finale. They fade because the labor of making the show became too heavy.

The editing.

Uploading.

Show notes.

And the endless feedback loops…

I'm my own best and worst case study of exactly this. 

I started my own show years back, went between formats, couldn't stay consistent, made up for time, changed formats again to try and get better, got tired of all of it, resorted to paying someone else to host it AND produce it, quickly got over that, went on break, came back from break, went to fully automated with my AI voice, and still.. have not published an episode for the last few months.

If you have a podcast right now, you know these feelings. 

😬 And honestly, you have three not-great options:

→ Do it yourself. That costs two to six hours an episode, every week, with no end in sight. And it starts after you are already tired from the part you love. It could go faster, sure, but you don't love the shortcuts. 

→ Hire an editor. That runs $75 to $300+ an episode, or $500 to $2,500 a month on retainer. And every episode usually needs a feedback call, your time, and your kind of over paying for it. Quality moves with the editor's workload, not yours.

→ Use a one-shot AI tool. Fast and cheap. But it gives you something that is not your show. A tool that has never heard your masters and leaves you doing all the assembly, all the decisions, and all the publishing. Nothing about it is truly intelligent, trained on what's working and watching out for you. Plus, it removes the human component completely. 

So many are stopping because they don't want to do these or pay for it anymore.

So after realizing I'm my own customer, I decided to fix it after proving it will work.

🔧 The Architecture

I designed the system in two halves.

The first half is The Studio.

It models a show from its public feed alone. 

We're talking it reverse-engineers the format to the split-second. The studio pulls the verbatim recurring scripts and measures the show's actual mastering sound rather than assuming one.

Chris masters at -20 LUFS, not the -16 that gets called “the podcast standard”. 

Forcing him to the standard would have made his show worse. 

The Studio catches that. 

It produces hooks cut from the body in the host's own voice, ranked titles, show notes, chapters, and mastering to that show's own numbers, all with a verification step after every stage so nothing moves forward unmeasured. And it learns over time, taking feedback, implementing and applying those lessons learned for your show as it evolves. 

The second half is The Production Farm.

This is where the scale lives. And quite honestly, what is SO excellent.

The Production Farm picks up jobs automatically, 24/7, always ready. 

It doesn't clock out and doesn't have a backlog. 

Whether I'm running 5 shows or 500, the infrastructure scales to meet it. Every job that comes in gets queued, claimed, processed, measured, qa'd and delivered, all without me touching it.

The thing with so much value here is not edited audio. It is a locked decision set that continually gets better with every revision. 

Editing is just the means. Everything I built serves that one moment: low-effort, high-confidence sign-off, repeated.

…here's the part I want you to really understand though.

After that overnight build, I had a working prototype. But I didn't have a product.

June 25th, I took the family away for an extended weekend. Val, the girls, a cabana, a pool. The whole point was to have fun and recharge.

With volleyball and summer school, this was our one weekend this summer to get away. 

But I had this thing sitting in my mind, playing on repeat. At that time it was me, Claude and a billion neurons firing with ideas. 

I found some quiet time at the pool and started workshopping the ideas on my phone.

Not to build.
Just to think. 

I played the whole concept out loud, what it needed to be, where it was weak, what the real product was underneath the prototype.

Yes. This was my cabana where I did the things I teach. Clearing my head, being with my thoughts and making a data-driven decision with confidence.

The business model was born from a place of data & clarity. I modeled onboarding, risk, margins, scale. EVERYTHING. And I was sold. 

When I got back, I spent the following week not just writing out a plan, I built it. 

Then I walked Chris through a complete review. 

He pushed on the right things. I used Fable to build out the rest and I am genuinely in love with it.

That's when it went from prototype >> a company.

But it was still missing something.

🤝 The Review Process

A few weeks in, I was talking through the product with Madeline, my EA.

She pushed me on something I hadn't fully solved:

• How does a host actually approve an episode? 
• How do they give feedback without it becoming a back-and-forth that kills the whole point? 
• How do you make the sign-off feel EASY?

The review layer came out of that. 

The reviewer's choices come back as a structured, brief, and precise enough for an agent to execute the revision without a human interpreting the feedback. 

Locked decisions carry forward into the next version. Every superseded version is archived as the institutional record.

A host should spend five minutes max on an episode and feel completely confident in what goes out.

It should feel like watching a Loom and giving comments.

That's the product.

And this is how fast it moved:

(I love this part because I'm really proud of it.)

How many tokens did I burn building this?

Over 4 billion.

I'm not saying that to flex - well, not entirely. I'm saying it because that number represents every iteration, every revision loop, every edge case I tested, every show I modeled before I was confident enough to show it to anyone and check all of MY boxes to launch a product with scale in mind from the start.

That is what it costs to build something you are actually proud of.

AND I STILL have not written a line of code. 

⚡️ Speed Is Not the Enemy of Quality

I want to address something.

Every week I get a version of this question: "You can either have it good, or you can have it fast."

Or: "Speed is fine for tech, but not for our industry."

This is a false dichotomy. And it is almost always posed by someone who lacks the horsepower to do both simultaneously.

Speed does not come at the cost of rigor. 

Almost all real work is done in intense, focused, deep sprints. 

No distractions. 
Limited meetings. 
Just real heads-down pushing.

Remember when Google Maps came on the scene?

It was slow, clunky, and an XML-driven mess. 

Brian Taylor was furious and rebuilt it from scratch in a 48-hour coding sprint over a weekend. It became the infrastructure that still runs today 

One-third the size, ten times faster.

The original iPod was conceived, designed, and shipped to customers within seven months of starting and became the foundation of a ten-year transformation that turned Apple into a trillion-dollar company.

The Empire State Building was built in 410 days in 1930. 

It has lasted a century. 

Compare that to the Millennium Tower in San Francisco. Five years, multiple committees, endless planning sessions. It is now leaning, deemed unsafe, and cannot be fully used.

Competence matters.

Speed and quality are not opposites. They are what happens when someone with real horsepower decides to stop making excuses and just build.

Casthouse🔗 went from a dare to a live system with 15+ shows in seven weeks. And that was with time to pause, iterate and me run my full business with this as a concept.

That is not recklessness. That is what it looks like when you solve one problem faster than anyone else.

📣 And It's Already Working

I'm working with multiple producers who are adopting this system for their shows and workflows. Including Chris. His show is now a Casthouse production. Go give it a listen. 🔗

Another one of them - someone who was in my mastermind - told me that before Casthouse, they were turning away paid jobs.

They couldn't take on more shows without adding headcount, and adding headcount meant adding risk.

Now they can say yes.

The Production Farm handles the load. They handle the relationships. 

That's the shift.

Finally…

💸 The Math Behind the Business

Chris did the market math out loud in that first conversation.

Between first and second-degree connections, the two of us already knew roughly a thousand people with podcasts.

A fraction of a tenth of a percent of all podcasts paying a thousand dollars a month to have their show edited automatically is a million dollars a month.

Chris was paying his own production team thousands a month at the time and had been telling them, with affection, that he could not keep doing it.

The Casthouse Founding pricing is:

→ $799 a month for 30 episodes a month 👀 for the first 30 podcasters

Every plan gives unlimited revisions, and every episode gets a full human listen. 

That's my commitment to keeping the human component to this.

Producers and agencies running 5 to 50+ client shows get a flat $299 per client show. They are the QC, it's their business! 

Their episodes run through the Production Farm fast, and go right back to their team. Better with each revision, and the new secret weapon to break through the unit economic blocks they're all facing now.

🙏 Why I'm Telling You All of This

I want to bring you along for this.

Not just the wins. 

The whole thing.

The midnight screenshots. The 4 billion+ tokens. The poolside Claude session. 

The conversation with Madeline that cracked the product open. 

The moment Chris listened to his own produced episode and said, "I can't believe what I saw. Any podcaster with an audio show can now edit their own show."

I am grateful. Genuinely.

I'm grateful for the dare that started this. 

I'm grateful for the people who pushed me to make it better. 

I'm grateful for every one of you who has followed along, asked questions, and trusted me enough to read this far.

And I'm grateful to announce that the person who dared me to build this in the first place...

...is now building it WITH me.

Chris Harder is joining Casthouse with me.

My mentor. 

My friend. 

The man who texted me a raw audio file and waited to see what I'd do with it.

We're doing this together.

And we're just getting started.

🎙️ If you're a podcaster

Click here to join the waitlist 🔗 and we'll process your show and give you a demo, at NO charge whatsoever. The proof is in the demo.

🤝 If you know a podcaster, please share this with them.

Your support means the world to me.