Why Layer Health and Investors believe it can solve the Scalability Challenge of Healthcare AI for a fresh $21 million

Artificial intelligence (AI) enjoys the moment as the hottest sector of venture investment, with over $100 billion pouring into the sector last year. In healthcare, AI accounted for 30% of all venture funds in 2024, and data shows 2025 is off to a strong start.

This momentum continues with the $21 million Series A announcement by Layer Health, an ambitious healthcare AI startup aimed at tackling some of the sector's worst issues and overcoming the biggest barriers to industry growth.

The round was led by Depine Ventures with participation from Flare Capital Partners, GV and Multicare Capital Partners. They are taking part in the cap table already containing general catalysts and inception health. This suggests the reliability of the company's approach.

Layer Health applies a large-scale language model (LLM) to perform data abstractions for medical chart reviews. At first glance, common and esoteric to outsiders, chart reviews are the fundamental tasks that underpin a wide range of clinical and administrative workflows within the health system (and for other ecosystem partners). They can involve answering very specific and rich questions through a vast amount of some of the most fragmented and complex data (medical records) in any industry.

When used to support clinical decision-making at the time of care, or due to management functions such as improved clinical documentation (CDI), chart reviews are still labor-intensive and highly technical. Depending on the use case, both structured and unstructured data should be scrutinized – should be interpreted with records, progress notes, imaging reports, lab results, and clinician-level understanding. On a large scale, this process is expensive and time-consuming. Especially since it is often done manually by highly trained professionals now.

These properties make chart reviews particularly suitable for AI. LLMS excels in processing, summarizing and interpreting unstructured data with speed and accuracy. Although LLMS had problems with “hastisation,” Layer Health argues that models trained with longitudinal data support output with cited evidence, helping end users to trust and validate the information presented.

Still, deploying LLM in real-world healthcare settings, especially across different clinical settings, is not an easy task. Emphasizing the flexibility of the Core AI platform and the ability to mitigate hallucination problems, Layer Health navigates complex and competitive markets. However, the deep experience of its founding team and a system-conscious approach to the unique challenges of healthcare institutions can help distinguish them.

Peel off the layer

While most high school students in the late 1990s focused on chatting on malls, Nintendo 64 consoles and Nokia mobile phones, Layer Health co-founders David Sontag and Stephen Horn were already debating how they might affect the world one day. Both were drawn to computer science and shared a powerful entrepreneurial drive.

Like many teenage friends, they ultimately pursued separate paths. Sontag received his Ph.D. He holds computer science and teaching positions held at New York University and Massachusetts Institute of Technology. Horng became a doctor and earned additional degrees in computer science and biomedical informatics. He is currently an emergency physician at Beth Israel DeConnes Medical Center and also leads the Machine Learning Initiative.

Both had promising and independent careers, but their desire to work together ultimately restored them. Everyday experience with HORNG's ER has provided direct insight into the complexity and inefficiency of healthcare workflows and data systems.

From the early 2010s, the pair began building test applications with Sontag students within the (at the time) homemade EHR in Beth Israel. Over time, they investigated a variety of AI use cases for both clinical and administrative teams, repeating them in many early models.

“We had originally deployed an algorithm to detect sepsis, but it was detected quickly. “After discovering that discovery early, we pivoted into the clinical workflow.”

As LLMS began to emerge as a transformative power for AI, the foundations of layer health began to crystallize. One of the first widely cited papers on the use of LLMS in healthcare was co-authored by Sontag and ultimate co-founder Monica Agrawal and former MIT student Monica Agrawal.

By 2022, the collective experience of Sontag, Horng, Agrawal and two former MIT students, Luke Murray (software engineer at Google and SpaceX) and Divya Gopinath (founding engineer at Snowflake Acquired Truera) led to the formal establishment of Layer Health.

Name layer

According to Sontag and Horng, abstraction of medical chart data lies at the heart of layer health AI platforms, but its modular architecture is key to the company's strategy. Each module supports specific features, but also contributes to and builds other features, allowing the system to learn and improve throughout the use case.

The company's initial focus is modules that support clinical registry reports, which are used to track results over time and support research, quality improvements and public health. This module has already been deployed in Froedtert & The College of Wisconsin Health Network and is used to abstract data in quality reports. According to Layer Health, AI has reduced the time required by “over 65%.”

From there, the layer plans to validate one of the following modules: This is real-time clinical decision support at point-of-care.

“The same chart review issues that are resolved in the clinical registry module are facing clinicians at the time of care,” Sontag said. “For example, one of the following modules will focus on real-time clinical decision support to help automate clinical care pathways, leading to more reliable and high quality care. This not only improves patient outcomes, but also leads to more timely and accurate revenue capture, quality improvements and research.”

Additional modules under development are intended to support hospital operations and revenue cycle management by enhancing the CDI and medical coding process. The broader vision, as the name suggests, is to provide an enterprise-level solution (a basic AI “layer”).

Chart reviews are more than just essential to providers. Life science companies and clinical research institutes also rely on it to answer very specific and nuanced questions, particularly when evaluating patients for clinical trial eligibility. Manually reviewing the charts and assessing thousands of patients against inclusion and exclusion criteria is slow and expensive, making it another ripe area for automation.

Layer Health recently signed a multi-year agreement with the American Cancer Society (ACS). Use this platform to extract clinical data from thousands of patient records related to studies, including Cancer Prevention Research-3. The transaction followed the success of pilots who accurately extracted real data in some time.

Overlaying competition and custom-made issues

Despite the promising early traction, Layer Health faces a critical battle in a competitive market within the industry that is difficult to expand. Healthcare systems struggle with people and process-related challenges that cannot be solved by technology alone. Even within the same organization, different departments may have their own configurations, workflows, and legacy systems that complicate implementations.

The idea of ​​transferable enterprise-wide AI solutions is appealing, but in reality there are still significant barriers. Layer Health acknowledges these complexities and believes its platform is designed to fill them head on.

“Many of the healthcare challenges are universal, but some are unique local. Our enterprise platform also allows hospitals to easily configure, evaluate and deploy AI for chart reviews on specific local issues. It is directly integrated with hospital electronic medical records and existing business intelligence platforms, and is already unavailable using AI Chart reviews to easily extend the hospital's existing workforce. Customers,” Sontag said.

Investors share this belief. Lynne Chou O'Keefe, founder and managing partner of Define Ventures, sees Layer's architecture as a key differentiator.

“Layer Health is designed to be a fundamental AI platform rather than a single-use AI tool. Many AI solutions in Healthcare are either very specific to a single workflow or require extensive customization for each customer,” says O'Keefe. “In contrast, Layer Health has built a generalizable LLM-based system that can interpret complex clinical data across multiple use cases. AI reasons across patient charts allow healthcare systems to derive clinician-level insights with minimal configuration.

Define Ventures had previously announced $460 million with two new funds, so that layer naturally fitted into investment papers.

“We believe that the most successful AI companies are companies that solve system-wide inefficiencies rather than providing surface-level automation. Layer Health embodies this paper by addressing the immense issues of clinical data abstraction and chart review. Organizations are naturally suited to our investment approach,” explains O'Keefe.

Flare Capital Partners also sees the value of the layer's low-friction deployment model and the potential to generate revenue for health systems that operate at tight margins.

“Layer Health's AI platform unveils the insights that generate strong revenues for health systems through its unique ability to unify clinical chart data with results. Driven AI breakthroughs, Layer Health can provide these insights with minimal integration and current costs. “This has made David and Team the foundations and trusted partners for all healthcare organizations that deploy AI.”

The final layer?

The goal of Layer Health is to become a binding AI organization across clinical, operational and research domains is ambitious. As early traction in clinical registries is reporting and expanding partnerships across providers and life sciences sectors, the company positions itself as more than an easy-to-use solution. However, the path to widespread adoption in healthcare requires not only technical strength, but also adaptability to deeply rooted workflows and fragmented infrastructure.

Supported by $21 million in fresh capital and investors betting on basic impacts, Layer Health is facing the following challenges: Its platform expands, delivers meaningful ROI and demonstrates that it can adapt to the complex realities of healthcare. If successful, the company not only distinguishes itself into a busy AI landscape, but also helps to define the extent to which key language models are integrated into the future of healthcare.