One guest filmed Tuesday afternoon. His episode was edited the next day.
Six episodes, seven hours of raw footage, three clients delivered. The gear was never the bottleneck. The editing was.
One guest filmed on a Tuesday afternoon. He had his edited episode the next day.
I film podcasts at my studio in Long Beach. Two cameras pointed at two people. Somebody then has to turn that into a finished episode. Cutting to whoever is talking, fixing the audio, delivering it to the client. Yesterday six full episodes got edited. Seven hours of footage. Three clients got their finished files, their galleries and their emails.
Nobody in that business delivers that fast.
23:39
The gear is not the problem
People always ask about the setup. The cameras, the mixer, the TV, the studio. I want to do a full tutorial on it at different budgets, because you can start a podcast with $500, and then you can go to $1,000 or $5,000 if you want to. But that part is easy.
At the beginning you get excited. You film all these podcasts, all these videos. Then you have five or ten of them sitting there and the editing is boring and it takes a long time. That is where it dies.
The bottleneck is always going to be the video editing and publishing that content.
6:17
Stop shopping for cameras. Count how many finished pieces you actually published last month, and fix that number first.
What the system actually does
Two cameras, plus the audio files coming from the mixer. Everything sits under one folder with the client name on it. One folder for the cam videos, one for the audio. That is the most important part.
Then I open the AI video editor and tell it, here are the two cameras from my client's podcast, start the multicam edit. It asks me where the cameras are and which client. It does the multicam and drops the finished file into the exports folder under that client's name.
From there the AI YouTube publisher takes the video the editor made and posts it. It writes the description, gives it the title, and creates the thumbnail.
There is a reason all that text is in the description. It is not only the algorithm. We want the client to appear in ChatGPT searches. Instead of going to Google, people are going to ask, what is the best smoothie place in Long Beach, what is a personal trainer near me, what is a laundromat close to me. There are bots always scrolling the internet looking for that information. If you make their life easier, they are more likely to take it. There is a lot of testing here, but we are setting up the foundation.
I did not write the code for any of this. That is vibe coding. Instead of me writing it, I have a conversation and say build this, and AI writes the code.
Put every shoot under one folder named after the client, with cam videos in one subfolder and the audio from the mixer in another.
Screen record your session while you work so you always have the footage to cut back to later.
Ask the agent to find the pauses and the parts where you were giving talking points, and make it give you a report before it cuts anything.
What it costs and what it saves
A multicam edit by hand, stacking the two cameras, takes around two to three hours per episode. I am fast. I can do it in about an hour or less. Six episodes is around 12 hours.
An editor is $150 to $300 per episode. Six episodes is $900 to $1,800. Then there is my own time. At $75 an hour, which is not my rate, 12 hours is $900. Call it a thousand dollars saved in two days.
29:10
That is not $900 sitting in my hand. It means 12 hours back. I can sleep better, which is part of being healthy. I can call more clients. I can shoot another episode.
What does it cost to run. Almost nothing. About a dollar a month for file storage, plus the electricity to run the laptop overnight.
Work out what 12 hours back is worth at your own rate before you decide a system like this is not worth building.
It still gets things wrong
Yesterday it sent a client a delivery page where the only way to watch the episode was a 15 gigabyte download. It checked the link. It never checked that a human could actually finish downloading it. The client emailed back saying it did not work. That was a mistake on my reputation. We fixed it within the hour and changed the process so it cannot repeat.
I spent a couple of hours going back and forth telling it what looked wrong. The camera sitting too long on one shot. An episode starting before the actual intro. A client getting confused. Every one of those became a permanent rule in the system.
I am the taste and the judgment. It is the hands that never get tired, and it works while I sleep.
24:08
Agents are AI employees
The simplest way to understand an agent is that you are hiring an AI employee. Same as an assistant in a company. Do this. Send this email. Organize my files. Pay this bill.
I have an agent that records my discovery calls, listens, gets me a transcript and drops a proposal for me to send back with the price and the next steps. I have an outreach agent that finds a category of business in Long Island, finds their emails and their socials, and writes the email for me. I have an AI mentor built from the public videos, posts and books of someone I respect, because I cannot call him and his workshops cost $3,000 in Las Vegas. I have one for finances, one for Facebook ads, and one that makes calls.
For Pops Padel in Freeport, the facility was still under construction. I generated the images with AI and sent the client a deck. They looked at it and said, I want this one and this one. Easy feedback, and AI takes care of the rest.
Every person who has a personal brand or runs a business can hire AI employees.
That is why it works on the client side too. They look at it and they see someone who knows what he is doing. The content creators around here are not doing this. If you are 23 and you show up with that, it gives you credibility before you say a word.
Send the visual before you send the invoice. Let the client point at what they like instead of imagining it.
Invest in the skill
Rich people invest in skills. You keep adapting to the new stuff instead of getting left behind. You learn how to shoot, how to run the business, how to use AI, and then you apply it your own way because you are a human with your own preferences.
I sucked at this. Terrible. I was frustrated. It took me over a year from the day I pictured AI editing my videos to the day it actually ran. A lot of mistakes, trial and error. It is still not perfect and it still makes mistakes. That is fine. I talk to the agent and say, fix it.
47:49
01
Buying better gear will not get more work published. Fixing the editing and publishing will.
02
Put a number on the hours you get back at your own rate. That is the real return, not the invoice you avoided.
03
Expect to suck for a while. It took me over a year of trial and error before this actually ran.
Got footage sitting on a hard drive?
If you have unused footage and no capacity to edit it, tell me what you are shooting and where it gets stuck. I will tell you what I would automate first.
Send me a text