.png)
By Kacie Geretz, Director of Growth Enablement
December 11, 2025
The original version of this blog appears on Libman Education’s website here and was published on December 8, 2025.
AI is everywhere today, and medical coding is no exception. Some people hear that and worry. Others are excited. Both reactions are valid.
Nearly every medical coding vendor is now claiming they offer some sort of “AI-powered” or “AI medical coding” solution. If you’re using pre-existing CAC technology, as most HIM professionals are, you’ve probably noticed vendors positioning their tools with these same terms.
Then come offerings labeled “AI-powered autonomous coding,” and the picture gets even murkier. Is this just CAC with different marketing? Is it AI that still depends on human review workflows? Or is it a different category of technology altogether? Without clear definitions for terms like “AI medical coding” and “autonomous medical coding,” it’s incredibly difficult to evaluate solutions or make confident decisions.
This article will cut through that noise by outlining the main levels of coding automation and explaining, at a high level, the underlying technologies behind truly autonomous medical coding so you can more clearly assess what vendors are really putting on the table.
This is not a new technology, and many of you are currently utilizing these in your organizations, however let’s level set and review for comparison sake.
CAC tools analyze clinical documentation and suggest appropriate ICD-10, CPT, and HCPCS codes, but human coders must review and validate every single code before anything goes to billing. CAC uses natural language processing (NLP) to scan documentation, identify key terms, and handle tasks like code lookup and organizing information. The final coding decision always stays with the coder.
Many CAC vendors now market themselves as “AI-powered medical coding” solutions. Technically, they’re right, since NLP is a form of AI. But this creates confusion when comparing CAC to fundamentally different technologies that actually finalize codes without human review.
This is a newer category that sits between CAC and autonomous medical coding, and to be honest, this is the category with the least consistent terminology and definitions. But I’ll give it a shot.
AI-powered medical coding solutions (you may also hear this category referred to as “AI-assisted medical coding”) leverage AI to assign codes to patient encounters, but some level of human intervention remains part of the standard process.
For example, an AI-powered coding solution may autonomously code a portion of the patient encounter, but require manual coding for the other portion of the encounter. These solutions also may require a separate platform or workflow, similarly to a CAC type of workflow.
Autonomous medical coding refers to any solution that assigns a complete set of medical codes, such as ICD-10-CM, CPT, and HCPCS, to a patient encounter and sends it to billing with absolutely zero human intervention. This applies only after all implementation steps and quality audit checks have been completed and approved for Go Live in Production.
The technology that powers true autonomous coding solutions is fundamentally different from one vendor to the next, with some relying heavily on machine learning and others relying on large language model/rules-based techniques.
This matters because the underlying technology can strongly dictate how effectively a solution performs across specialties and service lines. For example, solutions that rely heavily on machine learning may be better with smaller and more predictable code sets such as radiology, where patterns are consistent and documentation is structured. However, those same solutions may struggle in more complex areas such as the emergency department or surgery, where documentation styles and medical specialties are broader and less predictable (this is where a large language/rules-based engine would be better suited).
The role of coding professionals remains critically important in this highest tier of automation. The coding team’s expertise and deep knowledge of coding guidelines, payer rules, and organizational coding philosophies is essential for governing how the autonomous engine codes. Their insight is what ensures quality through ongoing auditing and what drives continuous improvement through structured feedback loops. Coders and auditors provide the subject matter expertise that helps the engine evolve, stay aligned with regulatory expectations, and reflect the organization’s own coding standards. At the same time, successful automation allows organizations to shift some coding staff into more strategic and complex work, such as inpatient cases, denial prevention, proactive auditing, and other higher value functions.
Autonomous does not mean replacement. It means augmentation. In practical terms, it means the repetitive work can be handled by technology so that the skilled coding and HIM professionals can focus on essential key functions such as governance, accuracy, compliance, and process optimization.
Choosing between CAC, AI-assisted coding, and truly autonomous coding isn’t just a technology decision – it sets the tone for your team’s first real experience with AI. When coding professionals understand what’s actually powering a solution and where it’s meant to be used, they’re better positioned to own how it’s governed, applied, audited, and improved over time. Most of the struggles I’ve seen weren’t about “bad” technology; they came from teams who were never given a clear view of how the tool worked or where it was (and wasn’t) designed to shine.
Kacie Geretz, RHIA, CPMA, CPC, CCA is the Director of Growth Enablement at Nym, where she aligns Nym’s product roadmap with the evolving needs of health system partners and serves as the externally-facing expert on Nym’s autonomous medical coding engine. A graduate of The Ohio State University’s Health Information Management program, Kacie brings deep expertise across the revenue cycle—having led revenue integrity programs, built managed care contracting and credentialing infrastructure, and driven denials and A/R process improvement initiatives. She is passionate about advancing healthcare automation and regularly shares insights on coding innovation and RCM transformation.
