In January 2023, Verily Life Sciences began to pull back from the many different areas of the healthcare sector in which it once played to better tailor its bets on becoming a more profitable company.
Once known for tackling everything from smart contact lenses and wearable sensors to surgical robotics and virtual clinics, the Alphabet company and Google sibling has since put down a slightly-more-centralized stated goal: to focus on products that enable precision in medicine.
“We are designing complexity out of Verily,” CEO Stephen Gillett wrote in a company blog post at the time, a missive that also brought layoffs affecting about 15% of employees.
Verily has continued to reshape itself, including into this year, when it let go about three dozen employees working on its immune system molecular profiling project, according to a March report from Business Insider.
The company has also seen turnover in its C-suite. Former Chief Medical Officer Amy Abernethy, M.D., Ph.D., stepped down last December to join a nonprofit, while former Carbon Health President Myoung Cha signed on as chief product officer in February, and ex-Microsoft leader Bharat Rajagopal joined in April as chief revenue officer.
Andrew Trister moved to Verily last August to be chief scientific officer after working as the deputy director of digital health and artificial intelligence at the Gates Foundation. Before that, he served as a founding member of Apple’s health team.
Fierce Medtech caught up to Trister at the Milken Institute’s Global Conference in Los Angeles, where he sat on a panel about the potential of decentralizing clinical trials, and had the chance to discuss where Verily is headed from here.
Conor Hale: I want to talk about the changes that Verily has been going through over the past year. There was a large reorganization that aimed to narrow its focus. There were some layoffs last year, which have rolled into this year. What can you tell me about where you see the company now and where it’s going?
Andrew Trister: Yes, I think if we were to roll back and just think about where Verily started—because this is the evolution of what’s been happening—Verily came out of Google X and began with a vision of impacting the cost of care, the quality of care and access to care across the entire ecosystem.
So there were a lot of stakes in the ground leading to the horizon, which we are going to move toward—however, we didn’t know how to get there exactly.
There were a lot of science projects. There was a lot of investment in terms of hardware design, plus partnerships with biotechs, pharma companies and medical device companies, and other explorations. Then, going into 2020 with the COVID-19 pandemic, that all shifted completely.
So a lot of that work, where things were already starting to coalesce—with the greatest example being the Dexcom G7 [continuous glucose monitor] which was a partnership with Dexcom—that all shifted. The dialogue in the company became how to move more toward how we can respond to these exigent circumstances—which I think was appropriate, and made a lot of sense. But in the times coming out of that, there was a recognition that we could not continue to deliver in COVID as a business. It’s not as much of a need any longer.
So what we really wanted to do was pull the basket together, and drive impact towards that original North Star—and that really began at the end of 2022 and has been going until now.
It’s probably going to be an 18-month journey for us. A number of projects, indeed, were stopped or shed, which did have impacts on the people that were working at Verily.
And, increasingly, what we find is that the infrastructure we built to support data ingestion from various sources—everything from health system records to claims data, and patient-generated data using wearables and other health devices—all of that should come into one harmonized location.
We call that the Precision Health Platform, and that has been a more than two-year project to build. And then will come the services we can build on top of those data, once they’re harmonized and tokenized appropriately, to support research and the ways that we care for people.
So now, still with that same North Star, we’re able to really focus on data, on insights from those data through machine learning and AI tools, and then the building of effector arms through our own services.
That’s similar to Onduo, which is a legacy program now rebranded as Lightpath, focusing on diabetes and cardiovascular disease, while recognizing that there could be third-party approaches as well in the same platform.
CH: I’m curious how much DNA you feel Verily still shares with Google.
AT: Tons. If you think about “organize the world’s data, and make it useful” as the credo for Google, just insert the word “health” in there.
So we’re going to organize health data and make it useful for people—it’s the same exact goal, but obviously it’s more regulated, and the utility may not come in at the individual level. A person may not know what to do with all this data, so we have to recognize that there are a lot of other actors that we have to bring along.
CH: Frankly, Google also has a reputation for taking on many different disparate projects and not completing all of them. I’m personally still mourning the loss of Google Reader.
AT: Yes, there are a lot of favorites like that.
CH: Verily has also had to make some tough choices. What has been the main criteria for what stays and what goes?
AT: I think that there are a number of things that we’ve aligned to. So some are that, when the projects began, there was nothing else in the space—so it really was greenfield, and our explorations could go in the direction we felt that could be really impactful.
But over the course of the time, from say the beginnings in 2015 to now, others have emerged as either leaders in that space or potentially showing that the impact is real. And rather than compete, we would say, “This is great.” We’ve been able to seed the market, or recognize that others have seen the same opportunity. So we really wanted to focus instead on what makes us distinctive and differentiated.
A good example of this is the work that we had begun in digital pathology, which has blown up. Now there are many, many companies in this space, so we didn’t feel that there was something truly different about our approach that would be viable in the marketplace given the competitive landscape. And so we pivoted away from working in domains like that.
Now we are in this Cambrian explosion of AI use cases … The use cases are vast right now because there’s so much need—it’s a representation of the market, but that doesn’t mean that everyone is going to survive.
CH: So how would you describe Verily’s core competencies and offerings?
AT: We continue to have a core competency in hardware design. We have a lot of excellent hardware engineers and we’re in a medical-device-focused area—so there’s a lot of the quality compliance systems and all of the regulatory pieces that surround making a medical device and not just a consumer device.
So, when we think about wearables and their impact in healthcare, it’s not just about the Fitbits or the Apple Watches or Garmins—our device, the Study Watch, allows us to understand the impact that these measures can have from a health lens, first and foremost. I think that is a really important differentiator for us and a core competency for the company.
Similarly, going back to your DNA question, we have a tremendous group that focuses on data science and machine learning approaches. And then, of course, we have an army of engineers who build things that can scale while still following quality and regulatory requirements.
The Precision Health Platform requires a lot of compliance measures. We just went through ISO 13485 [medical device certification] for the entire stack, for everything from data to the devices. There’s also 21 CFR Part 11 [electronic records] compliance, and Part 50 [human subjects protection] compliance. That’s really meaningful work when we think about how we support research and clinical affairs.
There were a lot of science projects.
CH: One in-the-weeds question about the Study Watch, or any other technologies at Verily. Are you going after the FDA’s [medical device development tools] qualification program for clinical trials, known as MDDT?
AT: We’ve definitely been following that space. The Apple Watch may have very narrow indications, but it’s excellent to see that come forward. I do think that we can offer something quite similar, and we’re excited to see how that process works.
CH: I also want to talk about Project Baseline. Back again to that DNA question, Project Baseline tackles a wide range of research—it’s not just cardiometabolic, it’s not just mental health, it’s almost everything. And they’ve come out with a lot of papers recently ranging from hand grip strength studies to obesity.
How do you see that fitting into the company’s overall vision? It may be described as a narrower focus, but it’s still taking on lots of various projects.
AT: I think that both leaning on our previous experiences, as well as the understanding of what Baseline did well and didn’t do well, I think we have an opportunity to really drive this concept of sponsored research from Verily—and to forward hypothesis-driven experiences for participants who decide to join in these studies.
Baseline, by and large, had an underpinning of tackling very important, fundamental questions about health and disease from a distributed, decentralized lens by leveraging technology.
But within that, there’s a sub-cohort in the Baseline Health Study that’s about 2,500 people—people who had very intensive, in-person visits at our clinics, and clinics at Stanford and Duke. Those are the people that we know quite a lot about, and we continue to publish on that data with our academic colleagues.
And the vision there is having healthy volunteers participate, so we can collect a whole bunch of medical samples and different kinds of remote measurements and so forth, to see what we can learn longitudinally. And that’s been ongoing now for seven years.
So we’ll see what happens with those participants and the study’s steering committee, but what I see is the potential to find pathways to answering questions that people have for themselves, and that we as a society have, that are going to be important.
We can focus in on a few hypotheses, and they could be big ones—such as what’s the impact of GLP-1s in cardiometabolic disease, as an example.
CH: I noticed that some of the papers published recently had about 2,400 people, and another had about 2,500—so it’s all the same group of people each time?
AT: Yes, they’re the same participants in every study.
CH: That’s really fascinating. Also, here at the Global Conference, there’s a lot of AI-focused panels tackling a lot of different issues. I want to ask you what you feel is central to keep in mind when we’re talking about translating AI into a regulated healthcare product?
AT: Yeah, we’re in an amazing moment. And for me personally, for the better part of two decades, I have felt as though there would be some coalescence of data, AI and healthcare, but I couldn’t put my fingers on what exactly that would look like.
And while wearing all sorts of hats, either during my residency, or even in medical school—where coming in as a computer scientist wasn’t exactly the normal path, so people kind of look askew at me like, “Why would you want to be here?”—I think that’s no longer the question. The watershed moment has occurred.
Many people are recognizing that AI will have a major impact on society—and it’s felt very slow going and then it happened all of a sudden in November 2022, with the release of ChatGPT. Now we are in this Cambrian explosion of AI use cases, and there’s a lot of excitement.
It’s probably going to be an 18-month journey for us. A number of projects, indeed, were stopped or shed, which did have impacts on the people that were working at Verily.
I think that Verily, again, has a deep DNA and understanding of what it takes to really raise the bar and go into this area—not headlong as a startup might, but instead to realize what it means to work in a regulated space. That means working with regulators at the [Office of the National Coordinator for Health IT (ONC)] and the FDA to think about what data actually inform.
Because some of these uses might be as simple as, how can I be motivated to go for a walk when it’s sunny outside? And that’s great—but a lot of this really does come down to things that are heavily regulated: Which drug should I take? How should I go through the health system? Should I even go?
And I think what we’re seeing in this space is, appropriately, efforts working on problems of decision burnout and the back end of healthcare. These can be things such as ambient listening and building a clinical record for a person.
So, yes, there’s a lot of hype right now around AI. Some of it’s warranted, but I do think we have to ensure that there’s going to be equitable access, with data that’s representative of people that may have historically been disenfranchised from participating in clinical studies or participating in the health system as a whole.
We saw that during COVID, where small fissures became chasms for people. So the promise here really is to address accessibility, cost and quality by leveraging AI tools.
But we really have to be thoughtful about how that comes about, and who all the stakeholders are, and bring them together. Our belief is that being patient-led and being equitable are going to be the main focus areas—while also recognizing that regulations ensure that they’re safe.
CH: You mentioned a Cambrian explosion of use cases. As the evolution continues, do you see any chances of mass extinctions along those lines?
AT: Absolutely, it’s inevitable in this area.
The use cases are vast right now because there’s so much need—it’s a representation of the market, but that doesn’t mean that everyone is going to survive.
The expectation I would have is that we would see lots of really interesting insights, which may be a little too early for the market, or might not capture the imaginations of people appropriately. Product-market fit matters here a great deal. And we’re going to find that regulators have been very far ahead of this, in a way that I applaud, both at the FDA and ONC.
I think, though, that there are risks in the startup community—where if they’re not aware of the way that the regulations are moving, they might be moving too quickly. And that’s always a risk that happens in spaces that are shifting, and this one is shifting exponentially every day.
CH: Where do you see AI and other technologies making the biggest gains in global public health?
AT: I was really excited about this while wearing my last hat at the Gates Foundation.
There are a number of things that are missing in global public health at the moment: One is that we don’t have digital infrastructure in a lot of the places where people obtain care.
The global majority of people obtain care in their communities, with many of them actually in their homes. And these are healthcare workers that may be volunteers or paid by the government who also live in the community.
And there’s been a massive effort over the last decade to digitize the work that those health workers do. But we haven’t yet built the loop to take the data that is obtained—so, is there a young woman of childbearing age who may be pregnant in a home? Or is there a young child with a fever that may represent having malaria or an enteric disease?
These kinds of questions have massive morbidity and mortality impacts globally, but the decision-making loop isn’t there yet to give the health worker levers they can pull, such as to give an antibiotic or send the patient to see a nurse or midwife.
So I believe that AI has the capacity to, in real time, provide insights that will improve the access and the quality of the care that is delivered in that domain—while recognizing that this is not just true for the Global South, this is true globally. We are seeing a massive crunch right now in the need for health workers.
So if we were to augment the health worker—whether that is a community health worker volunteer in Africa or South Asia, or the nurse practitioner here in the Global North—to help them make better quality decisions and work at the top of their license, then we may be able to avoid what will be an inevitable problem of the workforce’s supply not meeting the demand and people being unable to access care.