The Copenhagen-based healthcare AI company based Symphony on peer-reviewed research from the largest medical coding study of its kind, treating coding as a case of reasoning rather than a labeling problem. Now available via API.
Medical coding, the process of converting clinical records, diagnoses, and procedures into standardized alphanumeric codes used for billing, reporting, and public health information, is one of the most error-prone and ultimately administrative tasks in healthcare.
The American coding system alone, ICD-10-CM, contains 70,000 diagnosis codes. Mistakes are common, costly, and often invisible.
CortiA Copenhagen-based clinical AI company has developed a product specifically designed to fix this: Symphony for Medical Coding, an agent system that it claims outperforms OpenAI, Anthropic, Amazon, Oracle and Microsoft models by up to 25% on clinical accuracy benchmarks.
Available via API starting today.
The performance gap that Corti claims is based on a methodological difference. Most AI systems approach medical coding as a classification problem: predict the most likely code from a training distribution, given a clinical record.
The problem is that coding rules are constantly changing, making historically trained models structurally inadequate. Courtney’s approachDeveloped through a peer-reviewed framework, Code Like Humans, adopted at EMNLP 2025, one of the top machine learning conferences, it treats coding as a grounding task instead of coding.
“Most AI systems fail in medical coding because they treat it as labeling rather than reasoning. Correct coding depends on evidence, context, hierarchy, and interpretation of instructions. We built Symphony for Medical Coding to follow the same decision process used by expert coders, which is why the performance gap is so meaningful.” he said Lars Maaløe, PhD, CTO and co-founder of Corti.
The system uses four agents in sequence: an index navigator that extracts evidence that isolates conditions in the clinical record, an index navigator that searches the ICD alphabetic index for candidate codes, a chart validator that verifies candidates against guidelines, and a code matcher that sequences and validates the final result. Each step represents what a trained human coder would do.
The study was based on 1.8 million patient encounters, making it the largest peer-reviewed study of its kind.
The consequences of conventional coding are not only financial. Corti cites a peer-reviewed study of Danish patient data that found his system identified three times as many suicide attempts as were officially coded, cases that were present in clinical records and medication records but were missed by coders working under time pressure.
“Medical coding has been treated as a back-office cost center for decades. It’s not the data layer on which healthcare works. Getting it right changes what healthcare systems can see, decide and do,” he said Andreas Cleve, CEO and co-founder of Corti.
Without these cases, health systems cannot track trends, allocate resources, or develop effective interventions. The coding layer is not an administrative burden; health systems see themselves as such.
Symphony for Medical Coding is the first system Corti has built to work in both US coding environments, ICD-10-CM for diagnoses, ICD-10-PCS and CPT for procedures, and European coding environments without local retraining.
ICD-10 coverage for Europe, protected by WHO, is currently in beta as the company expands into the UK, Germany, France and Denmark. The system provides verifiable results: each assigned code is linked to the clinical evidence that supports it, and uncertainties are reviewed by a human.
It is available through the Corti Console, integrates with the Corti Agentic Framework, and supports both A2A and MCP standards. Enterprise and sovereign cloud deployments are also available.
Corti is founded in Copenhagen and also has offices in New York and London. It has raised $100 million in total and serves more than 100 million patients annually across health systems, including the NHS.
The Symphony launch is a commercial product built on Code Like Humans research, following Corti’s stated approach of validating ideas in peer-reviewed forums before turning them into production-level infrastructure.
