Abridge
Last reviewed
Sources
30 citations
Review status
Source-backed
Revision
v2 · 2,800 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Sources
30 citations
Review status
Source-backed
Revision
v2 · 2,800 words
Add missing citations, update stale details, or suggest a clearer explanation.
Abridge is an American healthcare artificial intelligence company that builds ambient clinical documentation software, the category often described as the AI medical scribe. Founded in 2018 in Pittsburgh, Pennsylvania by cardiologist Shiv Rao, Abridge makes an application that listens to the spoken conversation between a clinician and a patient, transcribes it, and uses generative AI to draft a structured clinical note that the clinician reviews and signs into the patient record. It is the most heavily funded company in clinical AI documentation, reaching a 5.3 billion dollar valuation after a 300 million dollar Series E led by Andreessen Horowitz in June 2025, and by mid-2026 it was live across more than 300 United States health systems supporting upward of 100 million clinical conversations per year. [1][2][26] Alongside Microsoft-owned Nuance, Abridge is generally regarded as a leader in the artificial intelligence in healthcare documentation market.
Abridge addresses a specific and widely cited problem in medicine: the administrative burden of clinical documentation. Physicians commonly spend one to two hours per day, much of it after clinic hours in what clinicians call pajama time, typing notes into the electronic health record. Abridge captures the visit conversation through a phone or desktop microphone, with patient awareness, and returns a draft note within seconds of the encounter ending. The clinician edits and approves the note rather than writing it from scratch. The company states that "over 90% of clinicians who start using Abridge continue using the platform in a meaningful way" and that the product reduces burnout by up to 60 to 70 percent. [1]
The company sells primarily to large enterprise health systems rather than individual doctors, and it differentiates itself through deep integration with Epic Systems, the dominant United States electronic health record vendor. By mid-2025 Abridge reported that its platform supported more than 50 million medical conversations per year across 55 clinical specialties and 28 languages, and that more than 90 percent of clinicians who begin using it keep using it. [1] The product has expanded beyond note drafting into medical coding and revenue cycle workflows, positioning Abridge as broader clinical infrastructure rather than a standalone transcription tool.
Abridge was incorporated in March 2018 in Pittsburgh. It is most closely associated with its co-founder and chief executive, Dr. Shiv Rao, a practicing cardiologist who trained at the University of Pittsburgh School of Medicine and completed a cardiology fellowship at the University of Pittsburgh Medical Center (UPMC). Rao treated patients at the UPMC Heart and Vascular Institute while also serving as an executive at UPMC Enterprises, the health system's venture investment arm, where he evaluated and backed healthcare startups. [18] That dual vantage point, as both a frustrated end user of the electronic health record and an investor, shaped the company's founding idea.
Rao co-founded Abridge with Sandeep Konam, an engineer and former product manager at UPMC Enterprises who became chief technology officer, and Florian Metze, a Carnegie Mellon University research professor specializing in speech. The company grew out of the Pittsburgh Health Data Alliance, a research collaboration among the University of Pittsburgh, UPMC, and Carnegie Mellon. [18] Abridge's first product was a free consumer mobile app that let patients record their own visits and receive a plain-language summary, with the goal of helping patients remember and act on what their doctor told them. The company later pivoted from this patient-facing tool to the enterprise clinician documentation platform that defines it today.
Abridge's technical leadership was strengthened in 2023 when Zachary Lipton, a Carnegie Mellon machine-learning professor and former Amazon AI researcher who had advised the company since 2019, took on the role of chief technology and science officer, leading its in-house AI research. [19] Rao has continued to see patients during his tenure as chief executive, a detail he frequently cites as central to the product's clinical credibility. As Rao put it when describing the company's design philosophy, "Every medical conversation is rich with the signals our healthcare system depends on. Abridge activates those signals in the background, silently handling the complexity so clinicians can focus on the human moments that matter." [1]
Abridge's system combines speech recognition with large language model summarization. A proprietary automatic speech recognition engine, tuned for medical vocabulary and difficult acoustic conditions such as background noise and multiple overlapping speakers, converts the audio of a visit into a transcript. The company emphasizes multilingual capability, including handling of code-switching and interpreter-mediated encounters across dozens of languages, a feature aimed at safety-net and community health settings.
The transcript is then processed by in-house language models that filter out small talk and produce a structured clinical note, typically organized in the standard SOAP format (Subjective, Objective, Assessment, Plan). Abridge has stressed that it develops its own models rather than relying solely on third-party systems such as those from OpenAI, which the company argues gives it more control over accuracy, customization, and responsible-use safeguards. A central design choice is linked evidence: statements in the generated note are traceable back to the specific moment in the source audio and transcript, so a clinician can verify any line and check for the fabricated content, or hallucination, that is a known risk of generative models.
In February 2025 Abridge introduced what it calls the Contextual Reasoning Engine, which extends the platform beyond narrative notes into the financial side of a visit. [7] It captures Hierarchical Condition Categories with supporting evidence, drafts ICD-10 diagnosis codes, and supports risk-adjustment models such as CMS-HCC Version 28, with the aim of reducing the manual back-and-forth between clinicians and billing teams.
In June 2026 Abridge disclosed that it is co-developing, with NVIDIA, what the two companies describe as the first foundation model built specifically for clinical conversations rather than a general large language model adapted for medicine. The model is trained on Abridge's de-identified clinical data using NVIDIA's Nemotron open model family on Blackwell infrastructure, with the stated goal of a system that "reasons clinically from its foundation" instead of being retrofitted for healthcare. [27][29]
The product's commercial moat rests heavily on its Epic Systems integration. Abridge launched a directly embedded experience called Abridge Inside and was named an early member of Epic's partner program for ambient documentation, letting clinicians work inside Epic's Haiku mobile app and Hyperdrive desktop interface without switching tools. Abridge has also collaborated with Epic and Mayo Clinic on an ambient documentation workflow for nurses built into Epic's inpatient nursing tools, and it offers a dedicated product, Abridge Inside for Emergency Medicine, that works with Epic's ASAP emergency department module. [17]
Abridge scaled rapidly across large United States health systems between 2024 and 2026, frequently moving from limited pilots to enterprise-wide deployment. By the time of its June 2025 Series E the company reported more than 150 enterprise health system customers, [1] and by June 2026 it reported being live across more than 300 health systems, including Northwestern Medicine, Emory Healthcare, and Johns Hopkins. [26]
Its largest publicly described deployment is at Kaiser Permanente, which rolled out Abridge across roughly 40 hospitals and more than 600 medical offices spanning eight states and Washington, D.C., making the tool available to more than 24,000 physicians. [12][13] Other named enterprise customers and partners include Johns Hopkins Medicine, which agreed to deploy the platform across about 6,700 clinicians, six hospitals, and 40 care locations; UPMC, the founders' home system, which is scaling toward more than 12,000 clinicians; [15] Mayo Clinic, which expanded use enterprise-wide starting with roughly 2,000 clinicians; [14] Duke Health, which signed an enterprise agreement covering about 5,000 clinicians across more than 150 clinics; [16] and Sutter Health, which deployed the tool to more than 1,000 clinicians. Additional health systems that have publicly adopted Abridge include Yale New Haven Health, UChicago Medicine, Emory Healthcare, Corewell Health, CHRISTUS Health, The University of Kansas Health System, UCI Health, the University of Vermont Health Network, Deaconess Health System, and Reid Health.
Health systems have reported operational benefits including reductions in documentation time and self-reported improvements in clinician work satisfaction and burnout, though most published figures come from the company and its customers rather than independent trials. Sutter Health, for example, reported in an early pilot that 78 percent of participating clinicians experienced improved work satisfaction.
Abridge raised capital at an unusually fast pace, with its valuation rising from roughly 850 million dollars in early 2024 to a reported 5.3 billion dollars by mid-2025. [2] The company's headline rounds are summarized below.
| Round | Date | Amount | Post-money valuation | Lead investor(s) |
|---|---|---|---|---|
| Series C | February 2024 | $150 million | About $850 million | Lightspeed Venture Partners, Redpoint Ventures |
| Series D | February 2025 | $250 million | $2.75 billion | Elad Gil and IVP |
| Series E | June 2025 | $300 million | $5.3 billion | Andreessen Horowitz, Khosla Ventures |
| Series E extension (reported, not officially confirmed) | April 2026 | About $316 million | About $5.3 billion | Andreessen Horowitz (reported) |
The 150 million dollar Series C of February 2024 was led by Lightspeed Venture Partners and co-led by Redpoint Ventures, with participation from strategic healthcare investors including CVS Health Ventures and Kaiser Permanente Ventures. [9][11] One year later, in February 2025, Abridge closed a 250 million dollar Series D at a 2.75 billion dollar valuation co-led by the investor Elad Gil and the venture firm IVP, with backers including Bessemer Venture Partners, CapitalG, Lightspeed, Redpoint, Spark Capital, CVS Health Ventures, and NVentures, the venture arm of NVIDIA. [6][7] This round notably did not match the lead investors named in some early reports; it was led by Gil and IVP rather than by Lightspeed and Redpoint, which were participants.
Just four months later, in June 2025, Abridge raised a 300 million dollar Series E led by Andreessen Horowitz with participation from Khosla Ventures, roughly doubling its valuation to a reported 5.3 billion dollars. [2][3] At the time, the company disclosed about 117 million dollars in contracted annual recurring revenue, a metric that counts signed recurring contracts including customers not yet fully onboarded. [2] David George of Andreessen Horowitz said of the round, "Abridge has both the vision and velocity to create lasting, systemic change, and we are thrilled to be part of their journey." [1] Through the Series E, Abridge had raised more than 750 million dollars in disclosed equity funding since its founding.
In April 2026, secondary-market and deal-tracking sources reported that Abridge had raised an additional Series E extension of roughly 316 million dollars, again associated with Andreessen Horowitz and reported at a valuation of about 5.3 billion dollars, bringing total funding to approximately 830 million dollars. [26] As of this writing the extension had not been formally announced by the company, and the figures came from market data providers rather than a direct company disclosure, so they should be treated as reported but unconfirmed.
In June 2026 Abridge announced a strategic equity investment from the drugmaker Eli Lilly together with the NVIDIA clinical-conversations foundation model collaboration, signaling an expansion beyond documentation toward what the company calls an AI-native clinician intelligence platform spanning billing, clinical decision support, payer adjudication, and pharmaceutical trial screening. [27][28] A new life-sciences module can surface clinical-trial eligibility from within the clinical conversation itself, a capability of direct interest to Lilly's drug-development pipeline. Abridge also disclosed a partnership with Artisight to combine hospital-room sensor data with ambient documentation across inpatient stays. [26]
Rao framed the strategy in terms of connecting the major stakeholders in care: "We've known all along we wanted to be able to connect the dots across the main stakeholders in healthcare, because the only thing that matters is business model innovation," adding, "Now the priority is how much impact can we create, and speed is everything." [26] By this point Abridge reported serving more than 250 million patients across its 300-plus health-system customers. [26]
Abridge is one of the clearest commercial successes of generative AI in medicine. Ambient documentation has become one of the first generative AI use cases adopted at scale inside large hospital systems, in part because it targets a concrete and expensive pain point, clinician administrative burden and burnout, without directly making diagnostic or treatment decisions. The company's growth, valuation, and customer roster made it a benchmark for the broader wave of clinical AI, and it was named to the Forbes AI 50 list for three consecutive years from 2024 through 2026 and ranked the No. 1 Best in KLAS for ambient AI in both 2025 and 2026. [20][30]
Abridge operates in a crowded and fast-moving field. Its principal incumbent competitor is Nuance, now owned by Microsoft, whose Dragon Copilot product (formerly marketed as DAX Copilot) carries deep distribution into hospitals through long-standing dictation contracts. A cohort of well-funded startups also competes directly, including Ambience Healthcare, which raised a large round in 2025 and emphasizes coding and revenue integrity; Suki, an AI clinical assistant founded in 2017; Nabla, a France-founded ambient documentation company; and DeepScribe. Epic itself has begun shipping native ambient and AI documentation features, a development that could pressure third-party vendors that depend on Epic integration.
The category also faces open questions that temper the enthusiasm. Generative models can hallucinate, raising patient-safety and liability concerns that drive Abridge's emphasis on linked, verifiable evidence; independent peer-reviewed evidence of clinical and financial impact remains limited relative to vendor-reported metrics; and reimbursement, privacy, and regulatory frameworks for AI-generated documentation are still maturing. Even so, Abridge's trajectory, from a free patient app born inside a Pittsburgh hospital to a multibillion-dollar platform embedded in the daily workflow of tens of thousands of clinicians, illustrates how quickly large language model technology moved into routine clinical practice.