两年内从初创到被收购：Grove AI 的临床试验变革之旅

Pharmaceutical companies often point to slow clinical trial enrollment as a chokepoint for drug development, but it’s no picnic for patients either. Physicians may be able to suggest a trial or two, but it’s hard for them to keep up with all of the drugs in clinical development, said Tran Le, co-founder of Grove AI. More often than not, patients are directed to conduct their own searches on a clunky federal database of clinical trials.

Grove AI gives patients an easier and more intuitive alternative: Just pick up the phone and talk.

“When we started two years ago and we pitched our simple voice call agent to the industry, people told us we were dreaming and we were crazy and that it would never happen,” Le said. “Now, we get trial protocols from [clinical trial] sponsors. And in these trial protocols, they write down exact tasks that the [AI] agent should do and that the humans should do.” Grove’s technology uses AI agents to automate many clinical trial recruitment tasks that have historically been handled by a human. These agents interact with patients, helping them find the clinical trial that’s a fit for their disease. Whereas a human recruiter recruits for one trial at a time, a Grove agent has information about many trials and can recruit for multiple studies. After forming in 2024, Grove raised $4.9 million in seed financing the following year. In January of this year, Hippocratic AI acquired Grove for an undisclosed sum. It’s a journey that began in the personal experiences of Grove’s founders.

Le and co-founder Sohit Gatiganti worked at Stanford Medicine, where they used their training as AI engineers to help the hospital manage various aspects of clinical trials conducted there. Le said the work showed them just how slow and manual clinical trial recruitment was. But Le’s perspective was also informed by her own experience as a patient trying to enroll in several clinical trials. She said she encountered clinical research coordinators who were inundated with thousands of inquiries. She also found herself poring through long, confusing documents about trial requirements.

When Le and Gatiganti launched Grove in 2024, AI technology had reached an inflection point with respect to voice capabilities. Large language models could generate outputs in real time and then convert those outputs to voice, Le said. That enabled the technology to take a clinical trial recruitment process that has been very document driven and turn it into a conversation. This capability helps both Grove and the patient, Gatiganti said. Many patients prefer to handle health matters by speaking, and this free and open communication enables Grove to gather more data on what trials may be a fit for the patient.

“It’s much easier to actually go through on the phone, like five questions, ask follow-ups, get everything sorted out in five minutes, versus sending someone a five-page PDF to fill out, sending it back and trying to go back and forth asynchronously,” Gatiganti said.

By Michael Blackman, MD, MBA Chief Medical Officer, Greenway Health®

Grove’s customers are pharma companies, contract research organizations, and clinical trial sites that use the technology under a software-as-a-service business model. The company works with these customers to identify the trials that are currently enrolling, capturing data about the studies in order to understand what questions to ask a patient.

Grove does not market its offerings directly to patients. Le said patients learn about the voice agent option through pharma company outreach to disease groups and clinical sites as well as online advertising. Patients who consent to receive more information may receive a call. Alternatively, the patient may initiate the call. Either way, the technology is transparent that it is an AI agent acting on behalf of the customer. While Grove’s initial agent was named Grace, it goes by different names depending on the pharma company.

If a patient is ineligible for any trials, the agent can send a notification when a suitable trial does start recruiting. Right now, the Grove agents are only available for trials in the U.S., but the company aims to expand to other countries in the future. Gatiganti noted that the agents are multilingual. Besides English, the languages covered include Spanish, Mandarin, and Vietnamese, and the company continues to add more. The agent can also switch languages in real time. For patients who prefer to tap on a screen, the technology supports text messages.

Despite encountering some initial skepticism, Gatiganti said the Grove technology proved itself by solving a real pharma company need: recruiting eligible patients to clinical trial sites quickly. Word spread among sites and sponsors. Hippocratic heard too. Le said Grove was not looking to be acquired, but Hippocratic approached the startup, seeing it as a fit for its growing business serving the life sciences sector. Both Le and Gatiganti are now general managers in Hippocratic’s newly established life sciences division.

Le sees Grove’s technology as part of a broader effort to make pharma companies “AI native.” While there’s a lot of industry focus on using AI for drug discovery, she said an AI native pharma company will apply these technologies across the continuum with humans and agents working together. Trial screening and recruitment is just one piece. Le said Grove’s future plans include building agents for the post-enrollment period to answer questions from patients and improve their clinical trial experience. Agents could help after the trial or even after a drug is approved. Beyond educating clinicians about a new drug, Le said agents might also help them navigate co-pays and prior authorizations.

“That is our vision right now,” she said. “It’s going to constantly evolve as AI models get better. But we think there’s a huge opportunity for the whole industry here to really take AI and accelerate to bring therapies to market faster.” Photo: -aniaostudio-, Getty Images