CPT, maintained by the American Medical Association, is the code set for procedures and services in US clinical billing. An office visit, a vaccination, a lab draw, a surgical procedure all have CPT codes. The code determines the payment from the insurer.
CPT and ICD-10 are paired in every claim. CPT describes what the provider did; ICD-10 describes why. A mismatch (a high-complexity procedure code paired with a low-acuity diagnosis) triggers payer scrutiny and, often, denial. Coding correctness is operationally important for any clinic's revenue cycle.
For a clinical AI, the CPT use cases are mostly downstream of the encounter: suggesting the visit level (99213 vs 99214 depending on time and complexity), surfacing the supporting documentation, flagging mismatches before the claim goes out. The AI should not be selecting CPT codes unsupervised; the regulatory environment around clinical billing is unforgiving and the AI does not carry the liability.
Vorel surfaces CPT suggestions with the supporting documentation, alongside mismatch warnings against the ICD-10 selection. The biller or provider remains the decision maker on every code that hits a claim.