Your Doctor Gets 18 Minutes. Netflix Gets 1,095 Hours. This CEO Is Fixing The Gap

You see the person responsible for your health for somewhere between 10 and 20 minutes, once or twice a year. Netflix gets three hours a day. One of those two relationships has enough data to know you. It is not the one keeping you alive.

Funding stage: Series A Plus. Lirio closed a Series A of just under $10 million led by Bill Hambrecht through Hambrecht Ducera Growth Ventures, then a $27 million Series B in September 2023, with total capital raised reported between roughly $59 million and $75 million, backed by Hambrecht Ducera Growth Ventures, Fine Day Ventures, Nixon Capital, Bon Secours Mercy Health, and Rochester Regional Health.

That asymmetry is the entire business case for Lirio, and Marten den Haring has been building against it since 2019. He is not a healthcare lifer. He is an economist by training, MSc and PhD, who spent 25 years shipping AI products for financial services, law enforcement, and national security: Oracle, OpenText, Chief Product Officer at Digital Reasoning in Nashville, then SVP Platform at Element AI in Montreal, the lab co-founded by Turing Award winner Yoshua Bengio. Six months into Canada, Nashville called him home, and Lirio was hiring.

He joined as Chief Product Officer in May 2019 to turn a Series A business plan into an actual product. The MVP went live around January 15, 2020. Do that math: six to eight weeks in the wild, and then the world closed. Lirio's product exists to drive people toward care. Care was shut, then reopened on 50 different state timetables into backlogs and staffing shortages, and nobody wanted more patient volume. Marten took the CEO chair in May 2021, in the middle of it, and describes 2021 through 2023 as a stretch where the team was openly asking whether product market fit simply was not there.

What flipped it was not healthcare. It was everything else. Consumers spent three years learning that they could get anything from their couch, and then brought that expectation to their care. Today Lirio delivers more than 400 million nudges a year and reports outcomes that are hard to argue with: nudges 4x more effective than standard health messaging, a 60% reactivation rate on patients lost for three or more years, 50% HbA1c improvement in under six months, and $4.4 million in direct margin added through targeted interventions.

The word “nudge” is doing a lot of work there, and Marten is careful about it. Ryan suggests it sounds like his wife on his fourth glass of wine. Marten's answer is the best line in the episode: if you want an automated version of your spouse, Lirio cannot help you. What Lirio actually does is treat every variable of a message as a decision, not the content alone: the channel, the hour, the frequency, the framing, and whether the right recipient is even the patient at all. Sometimes the highest leverage nudge goes to the daughter who manages her father's medications. And critically, Lirio does not ask anyone to download anything. Marten calls the platform headless: a brain that sits behind the CRM, the data warehouse, the marketing automation, and the engagement tools a health system already bought, and makes that existing stack work better.

The conversation lands, five days after launch, on ChatGPT Health. Marten's read is more generous than you would expect from a competitor. He thinks general purpose models genuinely help people find and apply knowledge, but he draws a hard line at memory and context, models will struggle to hold every prior interaction and every non-clinical detail that determines how you should communicate with a specific human. Then he offers the analogy that should worry everyone: in the mid 2000s, people had real privacy concerns about a rectangular search box called Google, and used it anyway, all day long, because it was convenient. We have a precedent. We know how this trade resolves.

On his own stack, Marten is direct about the tension every regulated founder is living. Lirio runs a comprehensive risk management framework, HITRUST CSF certification, SOC 2 Type II, and NIST, and it experiments with Cursor, OpenClaw, and frontier models inside that framework, because if you do not, the economics of building software stop working for you. His point is that guardrails are cultural before they are procedural: it is easy to make mistakes when nobody is looking, and in healthcare those mistakes have consequences.

Asked what he believes about the future that most people think is crazy, Marten notes that his worldview is less crazy than Elon Musk's, who had told The Economist five days earlier that humans likely lose control of AI within ten years. Marten's contrarian take: humans will still be in control in ten years, not by default, but because we will have seen what happens if we do not take charge. He does not want a world where a handful of people turn the rest of us into docile Labradors living at the pleasure of intelligent systems. He wants autonomy, relatedness, and human agency intact, and it is roughly where the episode turns into two guys negotiating for robot knees so Ryan can finally dunk a basketball.