Was Yoda Right About AI? with JJ Zhuang (Plus Pretty Good AI Episode: Pretty Good AI, One Year In)

Welcome back to Founder Mode!

I have two shows for you this week, and both feature my co-founder JJ Zhuang, so this is a bit of a special one.

First, check out the newest episode of Founder Mode. Then stick around for episode six of the Pretty Good AI Podcast.

🎧 Listen to Founder Mode Episode 76 here:

Was Yoda Right About AI? wit... Oct 1 · Founder Mode 33:21   ​

​

JJ is one of our first repeat guests on Founder Mode. The last time he was on, about a year and a half ago, he made ten predictions about where AI and software were headed. So Jason and I built a slide deck grading every single one against what actually happened, and made JJ sit through it live.

Some of the calls held up well. JJ said the best engineers would still matter because methodology beats syntax knowledge, and that held. He said "agent" would become 2025's most overhyped, ill-defined buzzword, and between vendors slapping the word on everything and Gartner predicting 40 percent of agentic projects get canceled by 2027, that one landed too. He also said AI would be overhyped in the short term but underestimated in the long term, which, looking back a year later, he admitted might have actually undersold how useful AI became.

Other predictions didn't hold up as cleanly. His call to hire fast-learning generalists over specialists graded out wrong at the market level, since most 2026 AI job listings still want deep domain experts. JJ's defense: that's true higher up the stack, at the model and infrastructure layers, but at the application layer, where Pretty Good AI operates, his own hiring bet turned out exactly right.

Final score: 6.5 out of 10.

Jason's take: that beats the average weatherman.

The second half of the episode flips the format. Instead of grading the past, Jason and I put JJ on the record about Pretty Good AI's future: the biggest bet we're making that other healthcare AI companies aren't, how you design an agent that's confident but wrong, where the line between human and AI work moves over the next year, and what a lean team has to do to compete against better-funded competitors with bigger marketing budgets.

🎧 Listen to Pretty Good AI Episode 6 here:

Pretty Good AI, One Year In... Sep 29 · Pretty Good AI Podcast 36:07   ​

​

This episode is different too. Instead of interviewing a customer, I sat down with JJ, my co-founder, to mark a year of selling Pretty Good AI to athenaOne practices and talk through how we actually got here.

A few things stood out. Our hiring approach favors curious generalists over specialists, including people straight out of college with zero experience, because AI now does enough of the specialized heavy lifting that curiosity and the ability to ask the right questions matter more than years in a role. Our pricing is usage-based and month-to-month, with no long-term contracts and no big upfront implementation fees, which forces us to earn every customer's business every month instead of locking them in. JJ's tagline for the company, "We Don't Sell AI, We Build Yours," captures the core idea: every customer ends up with something that looks almost like a completely different product, built around their specific workflow instead of a one-size-fits-all tool.

We also talked about compliance. Because we treat patient data as mission-critical from day one, certifications like SOC 2 and HITRUST end up being a natural outcome of how we already build, not a separate project bolted on afterward. And looking ahead, the next layer of our roadmap is referral management and prior authorization, two of the most manual, most time-consuming parts of running a practice today.

Here's how I summed up the conversation:

Final Thoughts

Both episodes this week are really about the same company from two different angles. Founder Mode graded how well we can actually see the future. Pretty Good AI showed what we built once that future arrived. JJ's year-old predictions about generalists, agents, and AI being underestimated turned out to describe our own company before we even realized it. A year in, the bet is paying off.

Watch the new Founder Mode episode, then listen to episode 6 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​