In Define Ventures’ AI investing thesis, released Thursday, the health tech venture fund outlines a vision of artificial intelligence readiness to reengineer healthcare. It’s investing in founders that are willing to do so.
Whereas AI was first used for isolated use cases in healthcare and life sciences, the technology is moving into an era of systemwide transformation. Define is investing in founders that have multiple products in the pipeline and can deploy them into multiple teams in an organization, Chirag Shah, partner at Define Ventures, told Fierce Healthcare in an interview.
Many health systems have taken first steps into AI, such as running pilots. Now, as providers face compounding economic pressures, AI needs to produce real results and cost savings. AI technology has advanced enough to be able to do that if guided by the right hand, Define writes.
Define Ventures manages about $800 million in assets, and its companies are at incubation, seed funding, series A and series B stages. The more than two dozen companies in its portfolio include 9amHealth, Cohere Health, Hims & Hers, Layer Health and Luminai.
“[Healthcare has] started with use cases that are fairly cut and dry, and, I would say, represent just the starting point in terms of the work that we can automate and take off people's plates,” Shah said. “We're starting to move further up the stack.”
To show how AI can rebuild healthcare, Define uses the analogy of a house, which it calls the House of Healthcare. Per the analogy, AI products impact healthcare in one of three ways—the front door, the foundation or the rooms—but understanding how all the components work together is key to achieving market hold, it says.
The front door of healthcare is the patient’s first contact with the system, whether a primary care visit or a Google search, Shah explained. AI is being leveraged to hyper-personalize outreach to patients.
Rather than mass marketing emails, healthcare organizations can leverage these "front door" AI tools to understand individuals’ scheduling constraints, medical history and social context to tailor the message, content, timing and method of outreach. The fund calls it atomic personalization.
The foundation, walls and hallways of the house are products that transform raw data into AI-ready data. Foundation models are essential for allowing other AI applications to run. Whereas data extraction and transformation used to be manual processes, AI has made those tasks significantly easier. It can also extract insights more rapidly.
The rooms are where AI touches the point of care. It could be clinical decision support, AI-developed pharmaceuticals, AI scribes or automatic prior authorization, for example. AI can automate processes to take work off providers’ plates and allow care to be delivered more seamlessly and efficiently.
“The power in this framework is in showing how these layers interact,” the investing thesis says. “Personalization at the front door relies on a strong foundation of data. Insight extraction in the hallways enables smarter automation in the rooms. Automation in turn frees up capacity for deeper personalization.”
Any of the three categories can be worthwhile to invest in, Shah explained, but one good product doesn’t make for a good company. Define seeks to work with founders that have visions for multiple products to expand their services within customer organizations.
Shah said Define invests in companies that can keep relationships with customers over a long period of time and expand their contracts rather than companies that have shallow relationships with many customers.
Finally, Define ensures the AI companies it invests in have a solid return on investment for customers, who are pressed to find cost savings.
“We make sure that, A, we understand how exactly those dollars are calculated. B, we go and validate that with the customers that they're selling into and C, we make sure that that's not the only thing that they're focused on. Ultimately what we can create defensibility here is that they've gotten multiple shots on goal per product, B, C and D.”