The Copenhagen-based health AI company built Symphony on peer-reviewed research from the largest medical coding study of its kind, treating coding as a reasoning task rather than a labelling problem. It’s available via API now.
Medical coding, the process of converting clinical notes, diagnoses, and procedures into standardised alphanumeric codes used for billing, reporting, and public health data, is one of healthcare’s most error-prone and consequential administrative tasks.
The American coding system alone, ICD-10-CM, contains 70,000 diagnosis codes. Errors are routine, expensive, and often invisible.
Corti, the Copenhagen-based clinical AI company, has built a product specifically designed to fix this: Symphony for Medical Coding, an agentic system it claims outperforms models from OpenAI, Anthropic, Amazon, Oracle, and Microsoft by up to 25% on clinical accuracy benchmarks.
It is available via API from today.
The performance gap Corti claims is grounded in a methodological distinction. Most AI systems approach medical coding as a classification problem: given a clinical note, predict the most likely code from the training distribution.
The problem is that coding guidelines change constantly, making historically trained models structurally inadequate. Corti’s approach, developed through a peer-reviewed framework called Code Like Humans, accepted at EMNLP 2025, one of machine learning’s top conferences, treats coding instead as a reasoning task.
“Most AI systems fall short in medical coding because they treat it as labeling, not reasoning. Correct coding depends on evidence, context, hierarchy, and guideline interpretation. We built Symphony for Medical Coding to follow the same decision process expert coders use, and that is why the performance gap is so meaningful,” said Lars Maaløe, PhD, CTO and co-founder of Corti.
The system uses four agents in sequence: an evidence extractor that isolates conditions in a clinical note, an index navigator that searches the ICD alphabetical index for candidate codes, a tabular validator that checks candidates against guidelines, and a code reconciler that sequences and validates the final output. Each step mirrors what a trained human coder does.
The research was based on 1.8 million patient encounters, making it the largest peer-reviewed study of its kind.
The consequences of conventional under-coding are not merely financial. Corti cites a peer-reviewed study of Danish patient data in which its system identified three times as many suicide attempts as had been officially coded, cases that were present in clinical notes 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 isn’t – it’s the data layer that healthcare runs on. Getting it right changes what health systems can see, decide, and do,” said Andreas Cleve, CEO and co-founder of Corti.
When those cases go uncounted, health systems cannot monitor trends, allocate resources, or design effective interventions. The coding layer is not administrative overhead; it is how health systems see themselves.
Symphony for Medical Coding is the first system Corti has built to operate across 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, maintained by the WHO, is currently in beta as the company expands into the UK, Germany, France, and Denmark. The system produces auditable outputs: each assigned code is linked to the clinical evidence that supports it, with ambiguities flagged for human review.
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 was 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 the commercial product built on the Code Like Humans research, following Corti’s stated approach of validating ideas in peer-reviewed forums before translating them into production-grade infrastructure.
