Subscribe | Advertise | PRO Member Hub

TOGETHER WITH KIN HEALTH

Perspectives from the people building the future of health AI…

We sat down with Arpan Parikh, MD, MDA, CEO and Co-Founder of Kin Health, to learn how the company is building a patient-facing AI platform around one of the most important moments in healthcare: the conversation between a doctor and a patient.

Arpan is a double board-certified psychiatrist and addiction medicine physician who spent years practicing medicine and building healthcare services companies. The idea for Kin came into focus when his mother-in-law was diagnosed with breast cancer. Despite having ten physicians in the immediate family, coordinating her care meant piecing together information from multiple doctors, EHRs, patient portals, and lab systems in a shared Apple Note. The experience reinforced a problem Arpan had seen throughout his clinical career: patients often struggle to understand, remember, and act on the care they receive.

Kin was founded by Arpan, his brother and fellow physician Amit Parikh MD, and Kyle Alwyn, who previously founded HeyDoctor, a consumer healthcare company acquired by GoodRx. GoodRx co-founders Doug Hirsch and Trevor Bezdek are also founding partners and executive chairmen of Kin.

The company's free app lets patients record appointments across specialties and care settings, then produces easy-to-read summaries, follow-up checklists, and information they can share with family members and caregivers. As patients use Kin across visits, those conversations begin to form a longitudinal personal health record. Kin is also developing the action and navigation layer to help patients take the next steps their doctors recommend.

In roughly four and a half months since launching, Kin had reached 36K+ users and recorded more than 17K medical appointments, identifying 140K+ downstream care actions in the process. The company recently raised a $9M seed round led by Maveron, with participation from Town Hall Ventures, Flex Capital, Eniac Ventures, Pear VC, and others.

Arpan shared why Kin started by building for older Americans rather than the typical early adopters of consumer AI, how the company thinks about clinical accuracy and privacy, why physician recommendations have become an important distribution channel, and how a free patient-facing product could eventually support a much broader healthcare navigation business.

Let's start from the top. What led you and your co-founders to start Kin Health?
I'm a physician by training, double board-certified in adult psychiatry and addiction medicine. I spent the early part of my career treating patients with complex substance use disorders and psychiatric conditions in New York City. I was also running a clinic, which opened my eyes to how difficult the healthcare system is for patients to navigate.

I believe we have some of the best scientific and clinical capabilities in the world in this country, but our outcomes don't always reflect that. A big part of the breakdown happens when patients have to understand a highly technical conversation, remember what was said, and follow through on the recommendations days or weeks later.

That became personal about four years ago when my mother-in-law was diagnosed with breast cancer. My wife and I became the quarterbacks for her care. Even with ten physicians in our immediate family, it was incredibly difficult to coordinate appointments, results, and recommendations across different health systems.

We ended up maintaining a shared Apple Note, manually copying information from MyChart, Athena, Labcorp, and other sources. We essentially had to build her personal health record ourselves.

After getting her through that journey, I kept asking why every patient couldn't have a simple tool that did this for them.

That's where Kin came from. We believe the physician-patient conversation is one of the richest and most underused sources of information in healthcare.

For readers hearing about Kin for the first time, what does the company do today?
The simplest way I describe Kin is Granola for patients.

It's an ambient AI notetaker that patients bring to their appointments, regardless of specialty or care setting. You open the app, press a big red record button, and have your appointment. When you're finished, Kin produces a summary at roughly a fifth-grade reading level, usually within 90 to 120 seconds.

We focus on three things: what did my doctor tell me, what do I need to remember and do next, and how can I share that information with the people involved in my care? That might be a spouse, an adult child, or a home health aide. We call that group your care circle.

In about four and a half months, we've reached 36K+ users and recorded more than 17K medical appointments.

Many of our most engaged users are Americans between 50 and 90 years old in smaller cities and towns across the country. Some have recorded as many as 60 appointments in 16 weeks. These are often people managing multiple chronic conditions who have been overlooked by much of consumer technology.

And Kin is completely free for patients, caregivers, and healthcare professionals who recommend it.

Healthcare has spent years building AI tools for clinicians. Why is now the right moment to build the equivalent for patients?
Three things have changed.

First is technology. Models are finally fast enough, clinically useful enough, and inexpensive enough to deliver an experience like Kin for free. I think we crossed that threshold around the fall of 2025.

Second is the consumerization of healthcare. Patients increasingly expect convenience, transparency, and a good experience alongside high-quality medical care. That expectation started with more affluent consumers in major cities, but it's spreading across the country.

Third is changing attitudes toward recording conversations.

I still practice medicine. A few years ago, if a patient had asked to record our appointment, I probably would have been uncomfortable with it. Today, I use an ambient scribe myself and ask patients for consent to record.

That creates a natural opening for patients to have the same capability. Physicians are already asking patients for permission to record their visits, and patients increasingly want to preserve those conversations for themselves.

Those shifts have made the timing possible.

You've said the physician-patient conversation should be the foundation of a personal health record. Why start there instead of traditional EHR data?
We actually tested that hypothesis before starting the company.

We used an off-the-shelf AI notetaker during appointments with friends and family, then pulled the corresponding clinical notes and records from EHRs and health information exchanges. We compared them side by side.

What became clear was how difficult it is to turn a traditional clinical note into something genuinely useful for a patient.

Those notes are written for clinicians, documentation requirements, coding, and other needs of the healthcare system. Ambient scribes have made it easier to generate detailed documentation, but they've also increased the volume of text patients have to make sense of.

The actual conversation often provides a much better starting point. It captures what the doctor explained, what the patient asked, which decisions were made, and what both people understood to be important.

Of course, conversations don't contain everything. Lab results, prescription details, and other clinical data may live elsewhere.

Our approach is to make conversational data the foundation, then use it to make sense of those other sources. Over time, that becomes a longitudinal personal health record controlled by the patient and portable across healthcare systems.

Walk us through the patient experience. How does Kin become more useful as someone records additional appointments?
The initial summary is useful on its own, but the experience becomes much more powerful once Kin has context across multiple visits.

We've already launched visit preparation. Say you've seen five doctors in the last month and have a cardiology appointment tomorrow. Kin can help identify questions to ask and important updates to share, like a medication change from your psychiatrist that could affect your blood pressure.

We also have chat, where patients can ask questions by text or voice based on what their doctors actually told them. If you can't remember the dietary changes your cardiologist recommended six weeks ago, Kin can find that guidance and reference the original conversation.

We've added tap-to-define for medical terms that can't easily be simplified, linking patients to clinically vetted sources. We also support Spanish, including appointments that move between English and Spanish.

The next layer is what we're calling the health graph, our longitudinal view of each user's healthcare information.

We're building one-pagers that turn that history into concise summaries for the patient, their caregivers, or a new physician. We're also working on Apple HealthKit integration to bring in additional clinical and biometric information, along with ways for Kin to participate in family text and group chats.

The goal is for all those individual appointments to become an increasingly useful record of someone's care.

You've been focused on older Americans and people managing complex conditions. How are you building for that population, and what's working on distribution?
I think a lot of consumer health companies make the mistake of building for people who look like their founders.

The person I think about is an 85-year-old grandfather in Lima, Ohio, who calls his diabetes "the sugar" and has never heard of ChatGPT.

He might have several chronic conditions, limited health literacy, and a fixed income. He's also someone who could benefit enormously from having a reliable record of what his doctors tell him.

We've learned that trust is probably the most important currency with these users.

One part is simplicity. We have an in-house design team focused on making Kin intuitive for people who may have very little experience with AI. Even the big red record button is intentional.

Another part is working alongside the doctors patients already trust. Someone managing congestive heart failure or cancer still needs their physicians, specialists, labs, and hospitals. We want Kin to make those relationships easier to navigate.

That approach has become an important distribution advantage.

We acquire users directly through paid, organic, and earned channels, but we're also seeing physicians, medical groups, ACOs, and MSOs recommend Kin to their patients. We're shipping materials to new practices every week so they can share Kin in waiting rooms and exam rooms.

The conversion when a doctor recommends Kin in person is orders of magnitude different from someone seeing a Facebook ad and downloading the app.

We're also in conversations with academic medical centers, health plans, employers, and benefits brokers about broader distribution.

You and Amit are both practicing physicians. How does that influence your approach to clinical accuracy, safety, and privacy?
I think lived clinical experience is incredibly valuable when you're building a product that touches medical care.

I've personally sat through tens of thousands of patient appointments. My brother and I have both continued practicing medicine, and we've built and operated healthcare services companies.

We wanted that expertise embedded in Kin rather than relying entirely on external medical advisors.

Our head of clinical AI, Rahul Dabathu, is a Stanford medical student and trained software engineer who leads our clinical validity work.

Every visit goes through a clinical quality pipeline. We're focused on two types of mistakes: errors of omission, where the system misses something important, and errors of inclusion, where the AI adds something that wasn't actually said.

We're also building a clinical review board where board-certified physicians will evaluate samples of de-identified summaries against the original recordings and transcripts.

Privacy is equally important. Patient-controlled consumer health products can operate under a different regulatory framework from traditional HIPAA-covered providers and health plans, so companies need to be deliberate about their safeguards.

Our approach is that users control their information. They choose who to share it with, can delete their data, and we don't sell it or use it to train models.

Clinical accuracy, user control, and simplicity are especially important for the population we're building for.

What does the technology stack look like under the hood? Are you building your own models or using frontier models?
We're generally using frontier models today, but Kyle built our architecture to be model-agnostic.

We can select different models for different parts of the workflow based on quality, speed, and cost.

Generating an appointment title is relatively simple. Turning a complicated medical conversation into an accurate clinical narrative and patient-facing summary requires much more sophisticated reasoning. We don't need to use the same model for both.

Today, Google's frontier models have provided the best overall combination of performance, speed, and cost for our user-facing outputs. But that can change quickly, which is why we don't want to be tied to one model provider.

We're also evaluating open-source models and the possibility of running some workloads on our own or partner hardware as we scale.

We've built supporting infrastructure ourselves, too. Our clinician and facility database draws on CMS data that we've cleaned and improved, while medical definitions link to vetted sources such as NIH resources and published literature.

The underlying question is how to deliver a clinically reliable, fast experience at a cost that allows Kin to remain free.

You've described the appointment summary as the starting point. How does Kin evolve from helping patients understand their care to helping them actually follow through?
This is where the opportunity gets really interesting.

Across more than 17K appointments recorded with Kin, we've identified over 140K next steps that patients were asked to take.

Those actions generally fall into three categories: diagnostics, including labs and imaging; medications, including prescriptions and over-the-counter recommendations; and specialty care, including referrals.

What's interesting is how little choice most patients realize they have.

If your doctor orders an MRI, you might automatically go to the imaging center affiliated with that health system. If you're referred to an endocrinologist, you might wait six months for the particular specialist your doctor recommended.

But maybe there's another specialist nearby who can see you tomorrow and has more experience with your condition.

Patients often don't know those alternatives exist or how to compare them on availability, convenience, cost, and quality.

We've already heard the recommendations directly from their physicians, so we have a very specific understanding of what each patient needs to do next.

We're building what we call an action orchestration layer to help people discover their options and complete those steps.

That can turn Kin from a record of what happened into a tool for navigating what happens next.

Kin is free for patients. How does the business model work, and how does the GoodRx connection influence your approach?
Kin has to be free for patients. That's foundational to what we're building.

The people who need it most may be on fixed incomes and managing expensive chronic conditions. A subscription would create a barrier for exactly those patients.

We think there's a more interesting commercial model in the downstream care actions that come out of appointments.

Because Kin captures what doctors recommend, we can help connect patients to the services they need, whether that's a lab, imaging center, pharmacy, or specialist.

One way to think about the action orchestration layer is as a highly context-aware healthcare marketplace. We know what a patient needs and can help them find options when they're ready to act.

We believe those connections can create a sustainable business while keeping the core product free. We've begun establishing the first commercial relationships and expect to launch the action-oriented product later in 2026 or early 2027.

Doug Hirsch and Trevor Bezdek, the co-founders of GoodRx, have been founding partners and executive chairmen from the beginning. Their experience has been incredibly useful in shaping our business model, partnerships, and distribution through healthcare professionals.

What are you seeing from early users? How are you measuring whether Kin improves understanding, adherence, or health outcomes?
We're still early, and now that we have enough usage data, we're beginning to measure those things more rigorously.

We want to understand whether Kin helps patients better understand their care, complete more of the steps their doctors recommend, and ultimately achieve better outcomes. We plan to publish that data.

What we can speak to today is the engagement and feedback we're seeing.

Some users have told us they're actually excited to go to their doctor's appointments because they get to use Kin. That's an unusual thing to hear, especially from people seeing doctors frequently for complex conditions.

We also ask users how they would feel if they suddenly lost access to Kin. The responses have been overwhelmingly strong, with people saying they can't imagine going back to managing their appointments without it.

That's encouraging from a product-market-fit perspective. The next step is demonstrating the effect on follow-through and outcomes with more rigorous evidence.

Five years from now, what does Kin look like if you succeed? And what do you think the industry still misunderstands about consumer health AI?
If we succeed, every American should be able to leave a medical appointment understanding what happened and what they need to do next.

Everyone should have a comprehensive, longitudinal personal health record in their pocket, with full control over who can see it. Caregivers should be able to stay informed about their loved ones' care, and patients should have real choice in how they fill prescriptions, schedule tests, and find specialists.

Ultimately, Kin becomes the primary interface through which patients understand, manage, navigate, and act on their healthcare.

I also think the industry needs to rethink who it's building for.

There's enormous attention going toward consumer AI products for a small population of sophisticated, affluent users. The bigger opportunity is helping Americans manage complicated, expensive conditions inside the traditional healthcare system.

Many of those people will continue relying on the physicians, hospitals, and specialists they already trust. We should spend much more time making that experience better.

More broadly, builders and investors need to start looking at healthcare through the patient's eyes. Patients should have much more control over their information, their decisions, and how they navigate care.

We need to stop trying to protect patients from themselves and start treating them like first-class citizens.

What stood out, what’s tricky, and why it matters…

Kin is the "Granola for patients," which is a useful way to understand the initial product. Patients record medical appointments, receive simple summaries and next steps, and can share that information with family members or caregivers. It's an intuitive starting point for a problem nearly everyone has experienced: leaving the doctor's office and struggling to remember everything that was discussed.

Healthcare has spent years building technology around physicians, from EHRs and ambient scribes to clinical decision support and workflow automation. Meanwhile, patients are left managing increasingly complicated care journeys across different doctors, health systems, and portals. For older Americans with multiple chronic conditions, keeping track of everything can be a real burden.

Kin's approach starts with the physician-patient conversation, which Arpan argues is one of healthcare's richest and most underused data sources. Those conversations capture what doctors explained, what patients asked, and what needs to happen next. As Kin accumulates visits and incorporates other clinical information, it aims to build a longitudinal personal health record that patients control and carry across providers.

The early traction is encouraging. Kin says it reached 36K+ users and recorded 17K+ medical appointments in roughly four and a half months. Some patients have recorded as many as 60 appointments in 16 weeks, with many of the most engaged users being older Americans managing complex conditions.

The longer-term opportunity is particularly interesting. Across those appointments, Kin has identified 140K+ downstream care actions, including prescriptions, diagnostic tests, and specialist referrals. Helping patients understand those recommendations is valuable. Helping them evaluate their options and complete those next steps could support a much larger healthcare navigation business, with a commercial model that keeps the core product free.

Kin is addressing a fundamental gap in healthcare: patients are expected to manage increasingly complicated care journeys without a reliable, accessible record of what their doctors actually told them.

The company has found an intuitive starting point in ambient documentation, with early adoption among the older, medically complex patients it wants to serve. Its physician-led approach to product development, patient-specific data strategy, and GoodRx-inspired distribution model give it several interesting advantages as it builds beyond the initial notetaker.

The next phase is really exciting. Kin wants to become a longitudinal personal health record and help patients navigate the decisions and actions that follow each appointment. That requires moving into more complicated territory around clinical reliability, data integration, patient trust, and commercial incentives.

If Kin can consistently help patients understand their care, keep caregivers informed, and complete more of what their doctors recommend, it has an opportunity to become an important patient-facing layer in healthcare. The strongest proof will come when that growing engagement translates into measurable improvements in how patients navigate and follow through on their care. We’re excited to see where Kin takes this and are rooting for the team as they work to make healthcare easier to navigate for patients everywhere.