Vivek Adarsh’s reasons for founding Mithrl are quite personal. About 15 years ago, his grandfather passed away from acute kidney disease. And while today there are approved treatments for the disease, 15 years ago there were no FDA-approved drugs specifically for the condition. Adarsh, a trained computer scientist, wanted to address this problem across multiple disease areas that still lack effective treatments today. Three years ago, he co-founded Mithrl, a company whose stated mission is to “help R&D teams go from idea to IND, 50% faster.”
As he explained in a conversation with GEN this week, “the question for me has been, what can technology do to accelerate getting medicines to patients a lot faster?” He and the Mithrl team believe that their artificial intelligence platform may be one of the keys. This week, the company announced a $20 million Series A funding round led by Obvious Ventures, with participation from Headline, AGI House, and several pharma executives, that will support the development of the second generation of Mithrl’s platform. The platform, dubbed Mithrl-1, comprises the company’s proprietary biomedical world model and an agentic harness for model routing, token optimization, and content orchestration. Mithrl is launching an early access program for the second-generation of the platform that will launch on September 21.
Mithrl’s platform combines validated biology curated from decades of peer-reviewed research as well as public and partnered datasets with agentic AI. Biopharmaceutical teams that deploy the company’s solution can combine that data, AI agents, and their proprietary data in their own environments. The platform cross-references data across disciplines to surface non-obvious connections using only published and verified data. Because the AI agents only reason over validated biology, rather than searching their way to answer from a broader pool of information, they require much less compute power to complete their tasks, according to the company.
That assertion is supported by results from a recent benchmark analysis, shared by Mithrl, that showed that running its infrastructure used 45% fewer tokens than standard workflows running the same frontier base models without customization. Finally, when the system generates hypotheses, it includes details about the sources used as well as a confidence score that captures what the system read, how it got there, and how much to trust its output.
Mithrl-1 is deployed in each client’s environment and harmonizes across existing frontier models. Each client’s implementation is then extended using their own proprietary data and pipelines. From there, teams can build and manage their own bespoke biomedical agents in-house. As new studies come in, Mithrl updates its knowledgebase and those changes are pushed out to customers on a regular basis giving them access to studies that are relevant to their work, and that could influence how they approach research questions.
These kinds of capabilities are important as AI systems become more embedded in biotech and biopharma labs. “Every player in this category is racing to generate more raw hypotheses, faster. But the industry is cracking under the weight of hypotheses it can’t triage or validate,” said Adarsh, who also serves as the company’s CEO. Mithrl set out to answer a different question. “Which of these will actually hold up in downstream experiments, and ultimately in patients? The platform doesn’t guess. It reasons from trusted studies and each client’s in-house evidence the way a client’s best scientists do, and it shows its work every time.”
To date, Mithrl’s platform has been used by several top-10 pharmaceutical companies, clinical-stage biotech companies, and genomics platform partners, according to the company. That list includes Elephas Biosciences, a company that develops an ex vivo tumor profiling platform. In April this year, the companies announced a scientific collaboration aimed at combining functional tumor profiling with AI-driven analysis to discover novel immunotherapy responses signals. The company could announce additional customer deals in Q4 of this year.
Some companies are already “AI-ready” which makes implementing Mithrl’s infrastructure more straightforward but most need some support to efficiently implement the company’s infrastructure, Adarsh says. “We have forward deployed scientists that go in, work with them, [help] them understand what knowledgebase does, and how we can make everything standardized.”
As the AI market matures, it has also become easier to make the case for implementing AI and communicating their platform’s benefits. A year ago, “you kind of had to work a little bit to educate where the value is” but that has changed in recent months, Adarsh noted. “I think there is a systemic level of education that has gone on over the past 12 months, mostly led by Frontier Labs, which is good for us.” There are also management changes happening across the biopharma industry and those shifts seem to be driving interest in onboarding AI, Adarsh said. He told GEN that he has noticed an uptick in conversations with company executives who “have a notion of the value that they are looking for” and are interested in understanding how Mithrl’s platform can help them get there.
So far the company’s tools seem to be making their mark on the industry. According to Mithrl, discoveries powered by its platform have contributed to over half a dozen customer-owned patent filings. It also claims that its biomedical world model delivers 16x more primary evidence per answer than frontier models alone and scored 0.96 for scientific correctness in expert-rated biomedical benchmarks, compared to 0.6 without its platform.
These proof points are helpful in a crowded market of AI-focused companies targeting drug discovery. Adarsh acknowledged the hype while discussing what sets Mithrl apart from other companies in the space. “Our north star is to be able to give scientific value and uniqueness to our partners,” he said. “What truly draws the revenue, and the topline for pharmaceutical companies, is uniqueness” specifically “patents.” He pointed to the platform’s contributions to multiple patents noting that customers have highlighted the competitive edge it gives. Those outcomes are “what differentiates us.”
And those outcomes, at least in part, are driving growing interest in the company’s platform. Adarsh told GEN that the current fundraising round was motivated to an extent by greater customer demand as well as a need to expand its capabilities. To the end, the new funding will support further development of the underlying biomedical world model as well as allow Mithrl add to its headcount, which currently stands at about 30 people. “In the order of priority, we have more demand to do deployments of our platform within big organizations, which obviously requires headcount,” Adarsh said. “Second is expansion across therapeutic areas.” The primary focus for now will be on disease areas that its target customer base works on including, immune-related diseases, oncology, diabetes, metabolic disease, and cardiovascular disease.
If all goes well, developing new medicines “won’t take 15 years” but “five years,” Adarsh said. “That is the personal motivation that led us to create Mithrl and that’s the path we’re on.”
