We recently connected with Alex and have shared our conversation below.

Alex, looking forward to hearing all of your stories today. One of the most important things small businesses can do, in our view, is to serve underserved communities that are ignored by giant corporations who often are just creating mass-market, one-size-fits-all solutions. Talk to us about how you serve an underserved community.

Medicare agencies operate in a broken economic model. Acquisition costs are high, regulations are complex, and retention is abysmal. The rational response has been to optimize for transaction volume: get the sale, move to the next lead. Post-enrollment support becomes a luxury only large operations can afford. For seniors, this means being sold a plan they don’t fully understand, then being left to navigate a labyrinth of benefits, copays, and coverage rules with no guide.

The underserved community isn’t just seniors—it’s also the independent agents who want to serve them well but lack infrastructure. We’ve worked with solo agents managing 2,500-member books who were on the verge of burnout, unable to provide the care they wanted to deliver. When your only resource is your own time, you’re forced to choose between serving existing members and acquiring new ones.

But the real vulnerability comes from what happens in that silence. Seniors enrolled in Medicare are constantly targeted—Facebook ads promising extra benefits, phone calls from competing brokers, mailers flooding their mailboxes daily. In the most insidious cases, this marketing is predatory: ads highlighting dental coverage, grocery cards, or vision benefits without mentioning that accessing them requires switching your entire Medicare plan. A senior sees “Get $900 in grocery benefits!” and doesn’t realize they’re being sold a plan change that might exclude their doctors or medications.

One story captures the cost of this perfectly: a 73-year-old woman who’d switched plans four times in a single Annual Enrollment Period. Each time, an agent called, highlighting some attractive benefit. Each time, she switched, thinking she was just adding coverage. When she finally needed care, she discovered her new plan didn’t cover her cardiologist or her primary medications. She’d been denied coverage for necessary procedures because she kept resetting her coverage during the year.

When we traced back her history, the pattern was clear. She couldn’t remember her original agent’s name—hadn’t heard from them in ten months. When competitors called, she had no way to distinguish them from her actual agency. Without a relationship or even basic brand recognition with her agency of record, she had no reason to stop shopping. Those agents who sold her plans weren’t stewards—they were opportunists chasing commissions. And her original agency, lacking the infrastructure for ongoing engagement, had inadvertently abandoned her to that cycle.

careCycle addresses this by making world-class post-enrollment care economically viable. We automate welcome calls, benefit explanations, appointment reminders, and renewal outreach while maintaining the personalization that builds trust. Agencies can serve every member consistently without expanding headcount. Seniors get reliable support when they need it, and critically, they build relationships with their agency that protect them from predatory switching.

The broader insight: underserved communities often exist because the economics don’t support serving them well. AI can fix those economics, but only if you build for the specific domain instead of bolting generic tools onto broken workflows.

Great, appreciate you sharing that with us. Before we ask you to share more of your insights, can you take a moment to introduce yourself and how you got to where you are today to our readers.

I’m Alex Doonanco, founder and CEO of careCycle. My entry into healthcare AI wasn’t planned—it was provoked.

In 2022, I was deploying voice AI systems across finance, legal, and healthcare, and I kept hitting the same wall: the technology worked, but it didn’t matter. Companies were automating the wrong things. They’d build a system that could handle 10,000 calls per day, but couldn’t remember why someone called yesterday. In regulated industries where relationships compound over time, that’s not innovation—it’s just expensive theater.

Medicare crystallized the problem. Agencies were losing 40-50% of members in the first year, not because of bad products, but because of silence. A senior enrolls, then hears nothing for eleven months until renewal. Meanwhile, they’re bombarded by competitors. When another agency calls, they can’t even remember their original agent’s name. The economic waste was staggering, but what bothered me more was the human cost—seniors left confused and underserved in a system designed to help them.

Before careCycle, I spent years in conversational AI watching companies chase generalization. Build one system that works everywhere, the thinking went. But “works” in this context meant “completes the call,” not “solves the problem.” In healthcare, that gap is enormous. A system that can schedule appointments but can’t navigate plan-specific coverage questions isn’t helpful—it’s a compliance liability that erodes trust.

We built careCycle around a different thesis: depth beats breadth in regulated markets. Our AI teams are specialists—one handles pre-enrollment qualification, another manages post-enrollment onboarding, and another focuses on renewals. Each maintains persistent memory, understands Medicare’s regulatory landscape, and integrates with agency workflows. This isn’t generic AI with a healthcare wrapper; it’s healthcare infrastructure that happens to use AI.

The results validate the approach. We’ve processed over 1.2 million member conversations with voice AI, and we’ve made our clients tens of millions in renewals revenue and retained revenue through increased retention. Most importantly, we’ve created something agencies trust enough to put between them and their most valuable asset—their member relationships.

What I want people to understand about careCycle is that we’re building the operating system for modern Medicare distribution. The opportunity isn’t to automate individual tasks—it’s to reimagine how agencies scale care. When a small independent agent can deliver the same level of consistent, personalized support as a large call center, they compete differently. When retention improves by double digits, the entire economic model shifts. That’s what gets me excited.

People spend a lot of time talking about how AI is going to transform the world as though it will be automatic. The reality is that no technological unlock can change the way we live without those bold enough to pour their lives into building those applications. LLMs alone will soon be powerful enough to power applications that make our world unrecognizable. But they won’t make any impact without being harnessed in thoughtfully crafted applications. That’s what we’re building at careCycle—not just better automation, but the application layer that turns possibility into reality for millions of seniors.

The metric I’m most proud of isn’t revenue or call volume—it’s what happens at the end of calls. When seniors say “thank you, sweetie, have a blessed day” to our AI, we’ve done something right. That warmth isn’t programmed; it’s earned through consistent, helpful service. That’s the future we’re building.

Let’s talk about resilience next – do you have a story you can share with us?

We’d spent months building, testing, and treating launch like something sacred. October rollout, two pilot agencies, controlled growth. Then September arrived and dismantled the plan entirely.

A mid-size FMO we’d been talking with called, and the voice on the other end had that specific timbre of controlled panic you only hear when someone’s staring at a collapsing operation. Their primary call center vendor—the backbone of their entire AEP operation—had just informed them they couldn’t handle the volume. Equipment failures, staffing shortages, something. The details didn’t matter. What mattered: Annual Enrollment Period started October 15th. They needed a replacement. Timeline: three weeks.

The rational answer was no. Every piece of startup doctrine said no. We weren’t ready for this. Technically, sure—the system could handle the load. But onboarding an operation that size, at that speed, with the entire Medicare industry’s most critical selling season about to begin? That wasn’t a risk; it was recklessness.

But I kept thinking about what “no” meant. They’d be forced to turn away thousands of seniors who needed enrollment help, or they’d cobble together a solution so compromised it would damage relationships they’d spent years building. And we’d launch safely in a few weeks with a nice case study and the knowledge that when it actually mattered, we’d protected ourselves.

We said yes. Not because we were confident—because the alternative felt worse.

What followed looked like controlled chaos from the outside and felt like barely-controlled panic from the inside. Eighteen-hour days integrating with their CRM, systems we’d never touched before. Training their team on workflows we were still refining. Customizing call flows for edge cases we hadn’t anticipated. Stress-testing at volumes we’d only theorized about. My co-founder and I effectively moved into their office. Sleep became this thing you did for a few hours before dawn, then ignored until the next night.

We went live October 12th. Three days before AEP. Three days before tens of thousands of seniors would start calling to enroll in coverage that would define their healthcare for the next year. If we failed, we’d fail spectacularly, publicly, in front of the entire industry.

The first week, we processed 47,000 calls. Caught issues in real-time—weird edge cases, integration hiccups, that one bizarre bug that only manifested when the system received three simultaneous calls from the same area code. Fixed them while calls were actively coming in. Delivered results that didn’t just match their previous vendor—exceeded them.

The resilience wasn’t in the execution, though that required everything we had. It was in the decision to take the risk when the easy choice was to wait. That moment defined careCycle in ways I didn’t fully understand at the time. We became the company that agencies called when they needed solutions that work under pressure, not just in demos. The company that would bet its entire reputation on an impossible timeline because the alternative meant seniors going underserved.

Looking back, we could have launched safely, built slowly, followed the playbook. Instead, we launched into chaos and proved something more valuable than any case study could demonstrate: when it matters most, we show up. That’s not a marketing message—it’s who we are.

We often hear about learning lessons – but just as important is unlearning lessons. Have you ever had to unlearn a lesson?

The lesson I had to unlearn was that understanding something intellectually means you’ve actually learned it. In entrepreneurship, knowing doesn’t count—only testing assumptions through real execution does.

I came from a world where you shipped at 80% and iterated to 95%. That final gap was expensive and usually unnecessary. I thought I understood that precision mattered in healthcare. I’d studied the domain, built the systems, felt prepared. Then we launched and reality taught me otherwise.

Our early metrics looked strong: 94% of calls handled perfectly, satisfaction scores high, efficiency up 40%. But at scale, that 6% gap wasn’t a rounding error—it was thousands of seniors getting information that was close but not quite right. One call stuck with me: our AI gave a technically accurate answer about drug coverage but missed critical formulary details. The agency caught it in QA, fixed it, no harm done. But I realized: in healthcare, “usually accurate” is useless. A 1% improvement at scale isn’t incremental—it’s the difference between hundreds of informed decisions and hundreds of decisions based on incomplete information.

The real learning came from what I couldn’t have known without shipping: businesses vote with their dollars, and agencies wouldn’t trust us with their member relationships unless we earned it through absolute precision. We rebuilt everything—implemented verification protocols, created “trust latency” where we optimize for accuracy over speed, embedded human-in-the-loop as core architecture, not a safety net. It made us slower and more expensive, but it made us trustworthy.

The goal isn’t to show what AI can do—it’s to show what it should do. I couldn’t have learned that distinction from research or planning. I learned it by taking the risk, shipping to customers, and uncovering the unknown unknowns that only execution reveals.

Contact Info:

• Website: https://www.carecycle.ai/
• Linkedin: https://www.linkedin.com/company/carecycle-yc-w25/