Women @ RopesTalk: From Lab to Launch—Donna Xia’s Mission to Revolutionize Insomnia Treatment

On this episode of Women @ RopesTalk hosted by health care partner Christine Moundas, IP transactions and licensing partner Megan Baca is joined by Donna Xia, founder and COO of a Silicon Valley startup revolutionizing insomnia care. Donna shares her unique career journey—from Harvard biomedicine PhD to McKinsey consultant, Sanofi dealmaker, and startup leader—and discusses the challenges and opportunities of bridging big pharma and fast-moving startups. The conversation explores Donna’s mission to expand access to cognitive behavioral therapy for insomnia, the innovative use of AI to scale clinical operations and improve patient care, and practical insights for adopting AI in healthcare and beyond. Tune in for an inspiring discussion on entrepreneurship, digital health, and the future of sleep care.

Transcript:

Christine Moundas: Welcome to Women @ RopesTalk, a podcast series brought to you by the Women’s Forum at Ropes & Gray. I’m Christine Moundas, a health care partner at Ropes & Gray based in New York and co-head of the firm’s digital health initiative. On this episode, I’m joined by my co-host of this series, Megan Baca, who’s based in Silicon Valley. Hi, Megan—before we get into the details about your guest, could you please reintroduce yourself?

Megan Baca: Hi, Christine, and thanks very much. I co-lead our IP transactions and licensing practice here at Ropes & Gray. I really focus on life sciences and technology at the intersection, and I work on a lot of AI-related licensing and collaboration transactions. So, to go with that, I lead our digital health initiative. And I represent pharmaceutical companies, biotech companies, and an array of other life sciences and healthcare companies in all of their complex transactions, which often involve licensing, and increasingly, artificial intelligence.

Christine Moundas: Who’s the guest that you’ll be interviewing on this episode?

Megan Baca: Today, I will be interviewing Donna Xia, the founder and chief operating officer of a Silicon Valley startup that focuses on insomnia care.

Christine Moundas: How did you meet and start working together?

Megan Baca: Donna and I first met when I was representing Iterative Health on various licensing projects and other related partnerships, and she was the SVP of corporate strategy and development there. Iterative Health is a really cool company that combines AI with a network of clinical research sites to help drug developers run faster, smarter, better trials and bring new therapies to patients faster, which is really exciting.

Christine Moundas: What would you say is most notable about Donna’s career?

Megan Baca: Donna is an incredibly brilliant and thoughtful person—I think you’ll see on this podcast. And what strikes me is how she has built a well-rounded perspective on her domain, with the ability to really see problems from many different angles, and it comes, I think, from her diverse background across different educational and career path opportunities. So, she did a PhD at Harvard in biomedicine, focusing on oncology care. After that, she became a McKinsey consultant, so no slouch. And she followed that by working at Sanofi in China on M&A and licensing, and equity investments. From there, pivoting to a smaller company, Iterative Health, where we met. And she’s now exploring the entrepreneurial world and has founded a new company in the insomnia space that she’ll tell us about in this podcast. So, she really has this incredibly rare and valuable perspective, crossing over research, consulting, big pharma, AI, and now, as a startup entrepreneur. So, I really think everyone will enjoy hearing her perspectives.

Christine Moundas: That’s great, I can’t wait to hear more. With that, I’ll turn it over to you and Donna.

Megan Baca: Welcome to the podcast, Donna, and thanks for joining us.

Donna Xia: Hi, Megan. Thank you for having me here.

Megan Baca: So, let’s just kick things off. Why don’t you go ahead and introduce yourself to our listeners—what your role is and a little bit about your background.

Donna Xia: Sure. Hi, everyone. My name is Donna. I’m currently a founding member of a startup in Silicon Valley, and what we focus on is building the largest insomnia network in the country, helping people to sleep better at night.

Megan Baca: So, Donna, whenever we meet up, one of the things that I always take note of is you have such an interesting career path, and I think you’re kind of a big-idea person as well. So, could you take us a bit back in time and give us a sense of your background, what you studied, and where your career path has taken you to get to where you are now?

Donna Xia: I started as a fake scientist at Harvard. I was trained as a PhD in biomedicine. At that time, I focused on oncology care. After that, I was thinking about how to scale innovation, because during grad school, you could only work on one project at a time, and also, translating that innovation to the day-to-day consumer space probably would take 10 years. So, after graduation, I was exploring more scalable ways to do that. One of the ideas I had by talking to Harvard alumni at that time was to find something in between business and science. As a transition, I thought McKinsey could be an amazing platform to launch at, so I joined McKinsey for two years, working with various clients across various different industries and functions. During that time, I discovered that my passion is in healthcare. Because of McKinsey’s platform, I did a lot of healthcare projects.

After McKinsey, I was thinking operation experiences would be also super helpful—hands-on operation experiences in addition to high-level strategy planning. So, at that time, there was this amazing opportunity with the Sanofi business development team in China where they were building the team to do anything from licensing to equity investments to merger acquisitions. I was just in the right place at the right time. I did probably 10 transaction deals at Sanofi in China, looking at innovation across the world, as well as in China. Actually, during that, I had the experience of working with smaller companies where I saw things moving really fast. I remember one example where we got a term sheet from biotech company within two days, and then, I think it probably took us two weeks at Sanofi to review, which is probably the normal speed in big pharma, but it’s just the contrast. Then, you realize, things are moving so fast in the smaller companies—I think that piqued my interest to move into the startup space.

In the beginning, I joined one of my best friends in Boston at Iterative Health. We were looking at AI-based diagnostic tools for colonoscopies, and I was the chief of staff there, basically leading all the operations in the company, except the tech space. And then, I think that’s when I dabbled in the startup world, and after that, I thought about starting my own company. So, that leads me to Silicon Valley with my best friends, two of them. And the reason I picked insomnia is I think diet and exercise have been recognized as the two important pillars for health, but I think there’s a missing pillar, which is sleep. I think there’s a lot of people who have difficulty falling asleep or staying asleep at night, and that’s what I am very passionate about to work on. And currently, we are building the largest insomnia care in the United States, so that’s something very exciting for me right now.

Megan Baca: That’s great. So, I do want to circle back to your current role, but before I do that, you have a really interesting vantage point, I think, on the issue that you raise, which is the distinction between how bigger companies, especially ones like pharmaceutical companies that live and breathe in a highly regulated, highly complex space, and smaller innovative companies that are looking to move really fast. And I understand from your background, coming from academia and then moving into business, that part of the goal is to get health technology and innovation to consumers and to patients faster, knowing that academia is not going to be necessarily the natural route for that. Now that you’ve probably done discussions and collaborations on both sides of the table, what strategies do you have to help them speak the same language and work together?

Donna Xia: I think that’s probably the most obvious question that I had when I was at Iterative Health, my first startup, and I think that’s where truly the advantage of both camps comes out. I think a smaller company does move so much faster in terms of R&D. Based on your transaction work, probably this is quite obvious to you as well, but I think the reason is, number one, the focus. I remember when I first joined Iterative Health, we only had one product, so the focus was very, very clear—that’s the priority for everyone. And I think in contrast, when I was at Sanofi, I was probably on 10+ projects. Then, if I’m talking about the VP above me, they were probably looking at 20 or even more projects at the same time. I think the sheer head space and time you dedicate to one project is dramatically different, and it allows you to move much faster in small companies.

Then, I think the second thing is the scale of the team, and then that also translates to the complexity of the process. At the time, the business development team at Iterative Health was only two people—just me and a manager—so, that allowed us to probably turn around a term sheet in one day, in terms of one review. But I think in bigger companies, there’s a rightful reason to do that, because many, many people are involved to make sure things are very balanced. I think for a complex project that’s really, really good, but it also means the trade-off we have to make is the speed. And I think for smaller companies, there is not that much coordination and process you need to go through. I remember at Sanofi, each term sheet needed to go through at least four rounds of reviews—finance reviews, business unit (BU) reviews, etc.—they are just optimized for different things.

And then, the last one is probably the pressure of survival. Smaller companies, there’s not that much runway, so I think the pressure to move fast is on top of everyone’s mind, versus, I think, in bigger companies, that’s probably less so, and being compliant with the regulations or other factor considerations is more important than the pressure to move fast. And as you alluded to and as I experienced, both organizations are optimized for different things. I think for the zero to one stage of a company (smaller company), it’s amazing because they can break things and move fast. But I think in terms of when you reach a certain scale, especially commercialization, the ability to operate a large organization definitely stands out. I think that one thing that speaks to that is the established order right now in terms of the smaller biotech doing the R&D and then bigger pharma doing the commercialization, so I think that probably speaks to the advantage of both in different stages of a company.

Megan Baca: That makes good sense, and also, I’m guessing is an area where you can add a lot of value, having seen both sides of the table so that you can speak their language when you are on the small company side and set expectations. But let me ask you now about your current focus of the work, your company focusing on insomnia. Tell us a little bit more about the network, the platform you’re developing, and what are the goals of the company in the next, say, two to five years?

Donna Xia: I think all the founders in the company, we all have insomnia. I’m a big fan of supporting health. In med school, we were all focused on diet and exercise, but I think sleep is such an important foundation that probably underlies every organ system in your body—our cardiovascular health, our metabolic health, the cognitive function—but it was just never up front. So, the three of us—I think one of them is also working for McKinsey and the other work for Google—we founded this network. At that time, when we got insomnia care, the first line of treatment is something called CBT-I, cognitive behavioral therapy for insomnia, but most people don’t get access to that. The reason is, in the whole United States, less than 600 people are trained in CBT-I, and for most of them, the wait time is six to 12 months. Not only that, most of them don’t take insurance given the large number of patients. So, we had this idea: What if we can train our own clinicians to do CBT-I? That will solve the problem at the root cause. So, we started the clinic a few years ago, and then started building our own robust training pipeline and hiring pipeline. Now, we’re probably close to 100 clinicians across the country and covering 50% of the population in the United States. And that’s the number of clinicians solving the problem. We use some of the AI tools there to help us to train clinicians. And in addition to the clinicians, another access issue I mentioned is the insurance, so that’s also where we spend lots of effort to make sure that our services are covered by insurance.

Megan Baca: That’s great. Talk about solving an unmet need. So, with 600 people nationwide trained in this, you’ve already increased the number of clinicians who are trained and able to serve patients in this space by 100 clinicians, and you facilitate patients finding those clinicians and then getting insurance reimbursement for their treatments. Is that the idea?

Donna Xia: Yes. I think we are very happy that we increased the supply of the clinicians by 15%, and then our goal is probably to further increase that at least another 15% by the end of the year.

Megan Baca: Wow, that’s incredible. And also a really fast roll-out. Like, how long have you guys been working on this?

Donna Xia: Three years now.

Megan Baca: Yes, that’s incredible. And you mentioned AI. I would like to spend a bit of time on AI, because it’s obviously the topic of the day but for very good reason. So, I’d love to hear about how you all are deploying AI in your business. And in businesses, I really see it as two sides of a coin where parts of AI use are actually deploying AI into the business model itself and into the business core operations and offerings, and then, the other side of it is using AI in terms of employee workflows and how you’re using it as like an efficiency multiplier for your team to get things done. So, could you tell us a bit about both of those aspects?

Donna Xia: There’s lots of AI work that’s probably native to the LLM model currently taking the majority of the attention in the market, like, tag, writing code, and then, there’s also a lot of contracting work as well. In healthcare, I think the majority of applications of that are probably note-taking. In terms of deploying that clinical decision-making, I think there’s definitely a lot of caution in that field—and that’s our approach as well for CBT-I, cognitive behavioral therapy for insomnia. The protocol was established decades ago—it has been demonstrated in hundreds of clinical trials, so I think it’s very structured and evidence-based. So, that part of the service we offer, we try to keep it as they are, so there was minimal AI involved in that to make sure there’s not many hallucinations in diagnosis and treatment. But we do use AI a lot in terms of supporting the delivery of that. For example, we do use AI in almost every operation process in the company. I can share statistics.

For our therapists, if they see patients for one session, typically it’s an hour per week. And our clinicians do very diligent work taking notes, which are required for the care but also required for the insurance payment. Typically, a good clinician spends 45 minutes documenting notes. More than a year ago, when we rolled out our own note-taking AI tools to all the clinicians, their note-taking time is now reduced to 10 minutes per session. I think not only that means now they can see more clients, but I also think the mental load that they have during the session is also significantly reduced. That means they can now focus more on the patient and the conversation. Once a clinician uses it, it’s 100% adoption. So, we have that in the clinic. That’s just one example, but then there’s other things. For example, we use AI in training our front office staff as well. All the transcripts will be screened by the LLM, and then there will be live coaching in terms of how they take a phone call from the patients, what questions to ask, and what information to gather—that’s also something that we wrote as well. And then, I remember talking to one of my friends in one of the largest call centers where they were saying in the past, it’s very hard to do call center staff coaching, because if you do sampling, you will have a biased sample, and if you cut across every call, it’s very hard to get granularity. But I think with the LLM, that’s just possible nowadays.

Also, almost every decision in the business is data driven. We track every single metric, and all the databases are connected. So, it allows us to probably deploy AI much more efficiently in the company. For example, in the front office right now, we do scheduling calls and all the administrative questions with patients. It’s all AI screened, and that makes sure the quality is there. And then, the rest is probably thinking about all the tedious work we can do using AI. Whenever there was a pattern to it, we used AI to do it, and even recently, I think we deployed AI to triage all the emails. So, now a clinician doesn’t need to remember 10 different emails, where to send their emails, what different questions—now it’s one single email, and then the AI will triage where the email goes to. And then, also for basic commonly asked questions, the AI will reply to the question as well. That will probably reduce like 50% of our email volume.

Megan Baca: That’s amazing. What specifically, if you don’t mind sharing, are the top few actual AI subscriptions and building blocks that go along with these solutions? What works so far, and have you had any that haven’t worked? Because a lot of this is trial and error in terms of how the building blocks of the different AI solutions can be deployed and used together to support your business.

Donna Xia: A majority of the team themselves are software engineers, so I think they all have subscriptions to all the major LLM models out there—Anthropic, Gemini, and Codex OpenAI. Maybe to just give an example—each one of them is probably good at doing different things, as you probably suspected. At least from my own experience, any contracting-related work I personally think Anthropic does a great job. And any medical research-related work, I personally love Gemini. And then, anything like writing-related, I think OpenAI is probably my favorite. But I think these things are evolving and changing all the time because the models are being released constantly. Another thing is, maybe because the teams are very comfortable with the AI themselves, we also use AI models beyond just the three LLM foundational models out there. For example, our note-taking assistant AI tools, we actually are not using any of those. We found a very mysterious AI model from Eastern Europe, but then it scored so high across all the metrics we wanted—accuracy, timing, etc. So, I think definitely besides the major models, we’re also looking at all the other models out there as well—it depends on what work we’re talking about.

Megan Baca: It just feels like practically a full-time job just keeping up with what AI is out there. And we have that at the firm, too. We’re constantly evaluating all of the legal-oriented tools as well, and it’s just a mind-bogglingly dense landscape of companies playing in these spaces. So, tell me, what is your, let’s say, two favorite uses of AI in your personal life, just for fun?

Donna Xia: Before I move on, I do want to say it’s so overwhelming to find all the tools as well. I don’t know whether it’s true in Ropes & Gray, but one of the things we recently started doing is everyone will huddle together once in a while, for example, on a weekly basis or something, or exchange what’s our own favorite AI tool and AI use cases, so that kind of cross-learning is spread across the organization as well. So, I think that helped as just a single person navigating the situation.

Megan Baca: Absolutely. Yes, that makes a lot of sense. Constant information sharing best practices is super useful.

Donna Xia: Yes. My favorite AI tool in personal life—I accidentally built an investment tool using Gemini late last year and early this year. I was blown away by the amazing analysis Gemini was able to put together and also the integration across different tools, like, Google Sheets, Google Search, etc. And I find that super helpful picking the stocks for myself, like, generating the portfolio analysis, reading through piles of annual reports, and I was blown away by that. That’s one. And then, the second thing I think is probably travel. I love travel and I think AI has made it so easy to do that.

Megan Baca: Same. I can’t possibly travel these days without AI being involved. So, last question. You probably work with a lot of AI natives, a lot of AI tech people. Like you said, your team’s very comfortable with AI, and so, it’s easy to forget that there’s a big swath of people—and some of them honestly in healthcare and life sciences—that have not really seen how AI can help them, help their companies, and help their jobs be more efficient and focus on the bigger picture, more important things and less mundane things. But what advice would you have for either the AI skeptics or the not yet adopters in terms of how you would suggest folks think about AI and entry points perhaps into making it more relevant for them?

Donna Xia: I think you being in Silicon Valley, and me, too, it’s easy to forget that we’re actually living in the center of the AI revolution, so we’re probably overexposed to AI. But I think travel definitely helps you to see maybe that’s not the case for everyone. I recently took a trip to Boston where I met up with my friends from Harvard Medical School, and they’re now in various parts of the sector of healthcare. I was probably seeing the contrast of how people are probably using AI much, much less. And probably a lot of them are very, very skeptical, as you said, which I think rightfully so—there’s a lot of early discussion about the hallucinations of AI. But I think the conversations I had with my friends was that you don’t have to make a big decision that I have to use AI whatsoever but more thinking about it as a sandbox—trying them out and seeing them as an experiment to see where they fit. And then another mindset I think that’s helpful is that every new technology always has problems, but I think that’s probably not where I want to focus. Maybe where I want to focus is where it could be helpful, because by definition of default settings, they are new, they will have problems. So, I think combine those two and try them out.

There’s also so many courses out there, posts out there, videos out there of people trying AI. That could be a good place to start, and once you try them, you will see the use of it. I’m definitely trying to showcase some of the AI uses. I remember one of my best friends from Harvard Medical School, he’s one of the leading researchers at Harvard right now. He was preparing a major presentation to one department in Johns Hopkins, and he was pretty skeptical once I convinced him to try AI out. And then, not only was the presentation itself significantly reduced in terms of one-week preparation to one day, but then, also the research part was amazing too—it reduced so much research literature. I think it’s just the first step is scary and also skeptical, but once people try under the mindset that it’s helpful for them, I think the rest is pretty natural.

Megan Baca: I totally agree. I think it’s kind of dipping your toe in and then seeing the power in certain areas and not letting the headlines or the naysaying examples get in the way of finding the broader swath of things that it’s actually really useful for, knowing that it’s not perfect and you have to be thoughtful about how you use it. Well, Donna, this has been delightful. Thank you so much for joining us—I really appreciate it, and we’ll hopefully work together soon.

Donna Xia: Yes, thank you, Megan. This has been so fun. Thank you so much for having me.

Christine Moundas: Megan and Donna, thank you both so much for that insightful discussion. And as always, thanks to our listeners. For more information about Ropes & Gray and our Women’s Forum, please visit www.ropesgray.com/women. You can also subscribe to this series wherever you typically listen to podcasts, including on Apple and Spotify. Thanks again for listening.