Ambience Healthcare, a U.S.-based artificial intelligence startup backed by OpenAI’s Startup Fund and top-tier venture firms, on Tuesday, announced the release of a new AI-powered medical coding model that outperforms board-certified doctors by 27% in identifying ICD-10 codes.
The model, built using OpenAI’s fine-tuning technology, marks what the company calls a significant step forward in automating the administrative burden in modern healthcare.
With this announcement, Ambience joins a growing wave of companies pushing artificial intelligence into the core of clinical practice, as the healthcare sector begins to embrace sweeping changes driven by automation. The timing of this development is notable: just weeks earlier, Saudi Arabia opened the world’s first fully AI-powered medical clinic, where an AI doctor named “Dr. Hua” leads patient consultations and treatment planning, while human physicians supervise from the background.
The Ambience model, while focused on coding rather than diagnosis, builds on this trend, reaffirming the potential of AI to transform not just how care is delivered—but how it’s documented, billed, and audited.
Outperforming Humans at the Administrative Core of Healthcare
The centerpiece of Ambience’s announcement is its new ICD-10 coding model, which the company says shows a 27% relative performance improvement over practicing physicians. ICD-10, short for the International Classification of Diseases (10th revision), is a universal coding system used by doctors, hospitals, and insurers worldwide. With more than 70,000 constantly updated diagnostic codes, the system is notoriously complex, making accurate documentation a difficult and time-consuming task for clinicians.
Ambience says its model is capable of listening to real-time patient encounters and automatically extracting the correct ICD-10 codes. The goal is to relieve doctors of the administrative weight of documentation and billing errors, which are frequent sources of inefficiency, frustration, and financial penalties in the healthcare industry.
“We’re not replacing doctors or coders,” said Brendan Fortuner, head of engineering at Ambience, in an interview with CNBC. “What we’re doing is we’re liberating them from administration, and we’re fixing mistakes that help make health care better, safer, more cost-effective.”
Rigorous Testing, Remarkable Results
To validate the model, Ambience developed a “gold panel” of coding labels agreed upon by seasoned clinicians who reviewed complex cases and determined the most accurate ICD-10 classifications. The company then recruited 18 board-certified physicians and tested them on the same set of cases. The AI model’s performance surpassed the average physician by 27% in coding accuracy.
“It shows for the first time that an AI system can actually surpass clinician experts at a very, very important administrative task, especially in coding,” Fortuner said.
Dr. Will Morris, Ambience’s chief medical officer and a former Cleveland Clinic executive, added that ICD-10 coding is more than just paperwork—it’s central to healthcare operations.
“It’s the cornerstone for quality,” Morris said. “If you think about it from a data perspective, it’s how you can compare and contrast clinician A to B, or health system A to B.”
Ambience’s technology is already being used by over 40 healthcare organizations, including Cleveland Clinic and UCSF Health. The startup has raised more than $100 million to date from investors such as Kleiner Perkins and Andreessen Horowitz. According to a report by The Information, the company is currently seeking new capital at a valuation exceeding $1 billion—a claim Ambience declined to comment on.
Built with OpenAI, Tuned for Healthcare
The model was built using OpenAI’s reinforcement fine-tuning tools, which enable companies to tailor high-performing large language models to highly specific domains—such as medicine. This specialization is crucial in healthcare, where accuracy, context, and clinical nuance can mean the difference between proper care and potential harm.
Beyond ICD-10, Ambience has also developed AI models capable of handling other medical codes such as CPT (Current Procedural Terminology), and is now exploring expansions into prior authorizations, utilization management, and clinical trial matching—areas notorious for bureaucratic delays and inefficiencies.
Fortuner noted that while the current focus is coding, the long-term vision is broader: building AI tools that can quietly but intelligently streamline the invisible but critical parts of medicine that often consume more time than the patient care itself.
The Bigger Picture Is AI Changing Healthcare
Artificial intelligence is no longer just a support tool in healthcare—it is becoming a lead actor.
Last month, Saudi Arabia recorded a breakthrough when it unveiled the world’s first fully AI-operated clinic. Unlike telehealth services or chatbot symptom checkers, Dr. Hua interacts with patients in real-time, collects clinical data, proposes treatment plans, and handles medical documentation. Human doctors are present, but only in a supervisory capacity—to review and approve decisions.
The AI clinic represents a radical rethinking of clinical workflows. And while it’s still a tightly controlled pilot, it offers a glimpse into how AI might take a central role in direct patient care in the near future.
Ambience’s approach is more conservative by comparison. Its tools operate behind the scenes—quietly listening, interpreting, coding—but its model’s performance, surpassing trained physicians, suggests that even the back-office tasks of healthcare are being reshaped by intelligent systems.
Ambience plans to roll out its new ICD-10 model to healthcare customers over the summer. If widely adopted, it could dramatically reduce the time clinicians spend on documentation, lower billing error rates, and standardize the way healthcare is coded and analyzed across systems.
