Being There at 2 A.M.: My Conversation on The AI Edge

A few weeks ago I sat down with Louis Fernandes for his show, The AI Edge. The episode went up this week: "He Built an AI That Helps Cancer Patients at 2 AM (And It's Changing Everything)."

The title refers to one part of the conversation, the chat agent on the Testicular Cancer Foundation website. We got to a lot more than that over the hour and change. Louis's audience is mostly commercial real estate operators and investors, so I expected to spend the time on Firmspace. We did some of that. But the conversation kept finding its way back to the question I find hardest to answer, which is how the pieces of my career fit together, and the honest answer runs through a pharmacy counter on Long Island, a small nonprofit in New York City, and my dad. Here's a recap for anyone who doesn't have 79 minutes.

The question I always get

Louis opened with it. You do a lot of things, Kenny. How do you usually explain that?

I gave him the long version. I'm an operations guy, a generalist. I run Firmspace, a premium private office company in Austin, Denver, Houston, and Atlanta. I run the Testicular Cancer Foundation. I'm CTO of a health tech company called Gryt. I've written three books. On paper it looks scattered, and for years I didn't have a clean way to tie it together.

The answer I landed on during the episode is trust and momentum. A lawyer closing a deal in one of our offices and a guy searching his symptoms at 2 a.m. are not having the same day, and I'd never pretend otherwise. But my job in both cases is identical. Someone is deciding whether to believe me, and I either make that easy or I don't. Then I keep the thing moving.

Where it started

In 2005, my senior year of high school, my dad was diagnosed with testicular cancer. He was 50, which is the wrong age for it. It usually shows up between 15 and 35, so it caught us completely off guard, and it pointed my career somewhere I hadn't planned.

I'd been a pharmacy technician since I was 15, first at the register, then behind the counter entering prescriptions and counting pills. I still tell teenagers looking for a job to go do that. You learn empathy and responsibility, and you learn how to talk to a scared patient before you're old enough to vote. After college I took an internship with a brain tumor survivor named Matthew Zachary at a small nonprofit called Stupid Cancer, and stayed about seven years. He and I were two sides of one brain, and we grew that thing well past what two people had any business growing. Ten years ago I came to Austin to run the Testicular Cancer Foundation, and Firmspace was born from the same group of people shortly after.

The 2 a.m. part

Now the part the title is about.

Testicular cancer is small as cancers go, about 10,000 cases a year in the US. A couple of years ago our website got maybe 100 visitors a month. After we took SEO seriously in the summer of 2024 and started pairing it with AI tools, that became 25,000 to 30,000 a month. A lot of those people arrive in the middle of the night, scared, typing things they wouldn't say to a doctor's face.

Google is a rough place to be at that hour. Search a symptom and you can talk yourself into a terminal diagnosis by sunrise. So we built the TC Navigator, a chat agent trained only on documents that real humans vetted. It isn't connected to the open web. It just knows what we know, and it's there whenever someone needs it.

Then I started noticing chat logs in languages I don't speak. That felt like an opportunity, so we translated the site into ten languages, and Google now indexes us in all of them. One person going through active treatment had a 186-message conversation with the agent, in Portuguese, probably somewhere in South America. I keep thinking about what care looked like wherever they were, and the fact that something we built in Austin was what they had at that hour. Not long after, a 50,000 Colombian peso donation showed up in Stripe, which is maybe ten or fifteen dollars. Someone found us at their worst moment and wanted to give something back.

We're not curing anything, and I said as much on the show. We like to say the doctor will cure you and TCF will heal you. Being available at 2 a.m. to someone who's terrified is the best thing I've pointed this technology at.

The parts where I look less smart

Louis asked what breaks when things get busy, and I didn't dress it up.

We turned on AI replies in the Firmspace support system, and because I ship things straight to production, the first live test was a paying member asking about a late fee. The bot's opening move: "Can you tell me more about who you are?" To an active member, in our own building. The fix wasn't a smarter model. It was context. You're talking to a member in Austin, private office, here's the history. Feed it that and it behaves. Our bot is named Ethan, and people now come to the front desk asking for him, because Ethan is so helpful. Ethan is not a person.

I also told the story of the email. I meant to ask our active Austin members about the battery status on their electronic door locks, and I sent it to the entire database instead. Present members, past members, cold leads going back a decade. Some people asked how I got their email. Others wrote back to say they missed us. And I copped to my habit of overbuilding, the Death Star app for a problem that needed a text message. Should I vibe code an RSVP app for a Fourth of July party? Probably not.

The lightning round

Louis closed with quick hits. Biggest mistake leaders are making right now: not adopting AI. Best professional decision I ever made: taking a risk on a tiny nonprofit as a new graduate. Most underrated system in any business: whatever your internal communication tool is.

The whole conversation was a good excuse to say out loud things I usually only think while I'm working at 1:30 in the morning, waiting for my son to wake up and tell me it's time for bed.

The episode is on YouTube here. The section on the navigator is the part I'd point you to first.