13 Agents Watching 76 Agents
I have two shows for you this week.
First, check out the newest episode of Founder Mode. Then stick around for episode two of the Pretty Good AI Podcast.
🎧 Listen to Founder Mode Episode 72 here:
On the latest Founder Mode episode, Jason Shafton and I get into how we price our services, what happens when AI quietly breaks instead of loudly failing, and why some of the biggest deals in healthcare tech are trading places right now.
We start with pricing. I walk through a deal I lost last week, not because our product was worse, but because the buyer wanted to pay for raw minutes instead of paying for the work getting done. Jason and I dig into why the old hourly-rate model doesn't hold up anymore, and why pricing close to what something actually costs to deliver is the safer bet in an AI world.
Then I share something I found by accident. I ran our bank statements and invoices through AI and caught more than 20 fraudulent charges that had somehow gone through, plus a pricing gap between customers who were paying the same rate for very different amounts of work.
The most surprising part of the episode is what I call the AI back office. I'm running 76 scheduled agents right now, and 13 of them exist purely to watch the other 63. Even with that many "supervisors," one of my agents quietly failed for eight days. It kept reporting success while writing its results to a folder that no longer existed.
I closed the episode with five takeaways, straight from the show:
- Decide the unit of value before you decide your price.
- Price as close as you can to what it actually costs you to deliver.
- Export your invoices and bank statements and let AI do the work. It's kind of insane what you can learn.
- Treat your agents like new team members. Someone needs to watch them and make sure they're actually doing what you asked.
- If you're good at writing and good at managing people, you have an edge in the AI era.
"Success, success, success is not the same thing as it worked."
-Kevin Henrikson
🎧 Listen to Pretty Good AI Episode 2 here:
My guest this episode is Jack Bretcher, COO of PartnerMD, a concierge medicine practice that's aiming to roughly triple in size. Jack runs both IT and operations, a combination most practices keep as two separate jobs.
Jack's rule for every acquisition is simple: don't break it. His team centralizes the easy things first, like billing, IT infrastructure, finance, and HR. But he won't force two practices onto the same system just for the sake of it. If it works fine running separately, he leaves it alone.
PartnerMD has chosen not to automate scheduling or add AI chatbots to the phones. Patients pay a premium for concierge care because they expect a live person on the other end, not a bot. When something is urgent, it rings straight to a live nurse or physician.
The practice tracks feedback in real time. Any score below a perfect 5 out of 5 gets flagged immediately to the practice manager, Jack's team, and the CEO. That real-time attention is a big reason PartnerMD finished last quarter with a 93 NPS.
Jack's take on AI is direct: it's a powerful tool when used narrowly and thoughtfully, but the idea that it can do everything is a problem. The number he checks every single morning is terminations, meaning patients who leave the practice, because it's the clearest signal of whether they're actually delivering value.
Here's how I summed up the conversation:
"You can automate almost anything except that patient experience. As Jack said, I pay you guys a lot of money to talk to people."
-Kevin Henrikson
Final Thoughts
Both episodes land on the same idea from different directions. Growth doesn't come from moving fast. It comes from knowing exactly what you're charging for, what you're automating, and what has to stay human. On Founder Mode, that means pricing around the real unit of value and watching your AI agents like new hires. On Pretty Good AI, that means centralizing the boring back-office work so your team can protect the one thing patients are actually paying for: a real person on the phone.
Watch the new Founder Mode episode, then listen to episode two of the Pretty Good AI Podcast. Subscribe so you don't miss what's next.
This podcast builds on the Founder Mode newsletter.
Let’s build.
-kevin
2810 N Church St #87205, Wilmington, DE 19802
Unsubscribe · Preferences