Today, we are pleased to announce our investment in Alterra Labs, a New York City-based company, led by Founder and CEO Akash Kapoor, building the evaluation and post-training infrastructure that healthcare AI needs to be trusted.
Flare Capital Partners is joining the company’s $12.5 million seed financing alongside our friends at Oak HC/FT, Company Ventures and a group of high-profile healthcare and technology angel investors.
The Origin Story
Our first meeting with Akash began with a frontier model failing in real time. We skipped the slides, and simply dropped into a simulated health system environment, with a live electronic health record, a coding workbench, and a payer portal. Akash set one of the world’s most capable AI models loose on revenue cycle work that takes a trained human hours to complete. Then we watched as the model cleared some tasks with ease, stumbled on others, and left behind a precise record of exactly where and why it went wrong. The results were eye opening.
Most importantly, the demonstration crystallized something we have been circling for a long time. Flare Capital Partners has spent the better part of a decade investing in applied healthcare AI, including companies such as SmarterDx, Layer Health, Suki and Qualified Health. Across that work, a consistent pattern has emerged: the capability of frontier models is advancing faster than the ability of health systems, payers, life sciences and AI builders to verify what those models can and cannot safely do. Akash and his team have built the missing piece.
Soon after meeting Akash, we moved quickly to join the seed financing, and we are honored to partner with the entire team.
Sizing the Opportunity
The center of gravity in AI development has shifted from pre-training on static data to post-training in interactive environments. Frontier labs now improve their models by having them practice realistic, multi-step tasks inside simulated software, scored by verifiers that confirm whether the tasks were accomplished appropriately. Reinforcement learning environments have become one of the most strategically important inputs to model capability, and the largest labs have reportedly considered spending in excess of $1 billion on environments and related infrastructure. The global reinforcement learning market is estimated to be worth roughly $11.6 billion and is rapidly growing.
Healthcare is where this shift matters most and where the infrastructure is least developed. Clinical and administrative work is long-horizon, tool-heavy, and consequential. A model that drafts a patient message, reconciles a chart, or works a denied claim operates across an Electronic Medical Record (EMR), a Picture Archiving and Communication System (PACS), a payer portal, and a dozen clinical conventions all at once. General purpose environments do not capture that complexity, and the benchmarks that exist today do not measure it. When we listened in on a frontier lab’s healthcare team during our diligence, the sentiment was unambiguous: “evaluation is a bottleneck across the board, and there is no credible benchmark for the workflows we care about.”
The same gap exists on the buyer side of healthcare. Health systems are being asked to govern a rapidly growing portfolio of AI tools, often with vendor assurances that are extremely challenging to fully validate. In fact, one large integrated delivery network we spoke with is hiring a dedicated team solely to evaluate AI for governance purposes, because as AI moves from drafting text to taking autonomous action, especially in patient-facing settings, the cost of an undetected failure rises sharply. Every participant in the healthcare AI ecosystem, from the labs training models to the systems deploying them, needs a rigorous, clinically grounded way to see how these systems fail before they reach a patient. That is a very good problem for a company to own.
Enter Alterra Labs. The company builds the full stack required to measure and improve how AI behaves in healthcare: simulation environments, graded evaluations, expert clinical judgment, and post-training data. Its environments are faithful and ground-truth reproductions of real healthcare software, spanning multiple EMRs, imaging systems, a payer portal, and voice, with tasks that span the routine, the difficult, and the edge cases that are a regular part of healthcare workflows. Each task comes with a clinician-authored rubric, a verifier, and a failure analysis that shows precisely how a given model went wrong.
What distinguishes Alterra Labs is the way it pairs people with infrastructure. The company employs an in-house team of more than 40 practicing clinicians who work alongside machine learning researchers to design environments, tasks, and rubrics within their areas of expertise. Multiple attending physicians review every rubric, and clinicians and researchers jointly review model trajectories to produce the final failure analyses. This is not data labeling. It is applied clinical science, and it produces a training and evaluation signal that general purpose data vendors cannot replicate.
The company serves three customer segments with the same underlying asset. Frontier AI labs use Alterra Labs’ environments and reinforcement learning data to improve model performance on high-value healthcare tasks. Applied AI companies use the same infrastructure to post-train open models toward frontier-level performance, on their own use cases, and on long-horizon workflows. Health systems, national specialty societies, and government bodies use it for independent evaluation of the AI they deploy and to set public standards for their fields.
Since its founding, the company has delivered calibrated sample sets to most of the leading frontier labs and has converted early engagements into paid partnerships with a handful of them. It has also established clinical and public health partnerships with hospitals, specialty societies and other healthcare organizations that give its work both domain authority and global reach.
The common thread across our diligence was execution speed. One advisor told us that every time he speaks with the team, something new has shipped. A former reinforcement learning infrastructure lead at a frontier lab ranked Akash among the strongest operators he has worked with and noted that the quality of the company’s data had improved steeply over a short period. In a market where the scarcest resource is researcher time, a partner that can deliver production-grade environments that most closely match real-world care, within weeks of a signed contract, is exactly what the labs are looking for. Combined with an obsessive focus on clinical quality, Alterra Labs is positioned to bridge the trust gap that is limiting healthcare AI adoption today.
The Road Ahead
The name Alterra comes from the Latin for “another earth.” It reflects the company’s view that the ground beneath healthcare is moving, and that the transition deserves clinical and scientific rigor rather than blind optimism. We share that view. The same environments that help a frontier lab improve its model can help a health system decide whether to deploy it, help a specialty society set a public standard for its field, and help a ministry of health evaluate tools before they reach patients in resource-constrained settings. Alterra Labs is building toward a position as the independent evaluator for healthcare AI, and the seed financing will support the expansion of its environments, its clinical expert network, and its commercial team across all three segments.
None of this would be possible without the clinical rigor and engineering velocity of this team. Akash’s path from machine learning researcher at the National Institutes of Health (NIH) to Columbia Medical School student to founder, informed by his own experience as a patient, gives him a rare command of both the science and the stakes of healthcare AI. His leadership is complemented by experienced machine learning researchers, a clinical leadership group that includes practicing physicians serving as experts in residence, and an experienced operations team.
Flare Capital Partners is excited to collaborate with the Alterra Labs team alongside our friends at Oak HC/FT and Company Ventures as we work to make healthcare AI measurably safer and more capable. To learn more about Alterra Labs or our other investments, please reach out to Parth Desai (parth@flarecapital.com), Alyssa Tsenter (alyssa@flarecapital.com), or any member of the Flare Capital Partners team.
Read the original announcement in Fortune here.
Stay in touch with Flare Capital
About Flare Capital Partners
Flare Capital Partners is a leading healthcare technology venture capital firm advancing innovation-driven companies to improve positive health outcomes, broaden care access, and lower healthcare costs. We partner with exceptional founders solving healthcare’s hardest challenges, supporting each with our deep sector expertise, unparalleled industry resources, and proven access to commercial opportunities. Our team of established investors and senior operating executives has invested in 80+ companies and has nearly $1 billion in assets under management. Learn more at www.flarecapital.com.
