Nabla is a clinical-documentation AI platform with documented API and EHR-integration paths for transcription, note generation, dictation, coding support, and FHIR-compliant normalization of Nabla-generated notes. The public materials describe both server-side APIs and embedded EHR workflows, with several functions gated by API date or product route. [1][3][8]

Documented product functions

The Core API introduction describes medical encounter transcription from raw audio, structured clinical note generation from transcripts, edit-with-instructions for Nabla-generated notes, medical dictation, Dot Phrase suggestions, a custom dictionary, and a feedback reporting path. The same page says these capabilities are intended for clinical workflows and integration into telehealth platforms, EHRs, or voice-enabled documentation tools. [1]

The documentation for normalized data generation says Nabla can extract ICD-10 and LOINC codes from a Nabla-generated clinical note and return conditions and observations in a FHIR-compliant structure. It does not describe general extraction from arbitrary source inputs. The request body includes a note object and an optional flag for returning corresponding note-problem titles when an Assessment & Plan section exists in the input note. [4]

Workflow and integration settings

Nabla’s EHR pages describe two implementation paths. One is a front-end integration through Nabla Connect, described for desktop-based EHRs and framed as a simple flow that can use Nabla’s existing user interface. The other is a back-end integration through the Core API for deeper UI integration across EHRs, including mobile-app contexts. [8]

The EHR and Epic pages describe encounter context moving from the chart into Nabla and completed output moving back into the chart. Documented outputs include notes, patient instructions, problems, vitals, diagnoses, and code suggestions, with Epic-specific descriptions for desktop, mobile, and telehealth use cases. [3][7]

Useful implementation detail

One specific boundary is the API date behavior around normalized data. The changelog says the newer ICD-10-CM 2027 code set is available on API version 2026-07-01 and later, and the 2026-10-05 changelog entry describes the newer normalization engine as using that code set. Implementers should read that as version-scoped behavior, not as a claim that every prior version received the same code set automatically. [2][6]

Versioning, scope, and documented boundaries

The changelog says Nabla releases dated API versions for stability and predictability, so integrations are not affected unless a newer version is explicitly adopted. It also separates date-based behavior from non-breaking updates such as model improvements and bug fixes. [2][6]

Several features are versioned or route-specific. The Core API introduction states that real-time and file transcription endpoint names changed at a specific date, while asynchronous dictation, custom dictionary support, and edit-with-instructions each apply from specific dates. The changelog also notes that asynchronous normalized data generation and related endpoints were added later than the base note-normalization route. [1][2][6]

Operational context and organizational use

Nabla’s public pages position the software inside clinician workflows rather than as a separate charting tool. The EHR pages describe documentation, dictation, and coding flowing into the chart, while the connect page describes an embedded encounter flow that opens a session URL inside an iframe and then exports the completed note and patient instructions back into the EHR. [3][5]

Vendor-facing pages also describe enterprise administration features such as customizable retention policies, SSO / SAML / SCIM integration, SMART on FHIR login support, and healthcare-grade infrastructure. Those claims appear on marketing pages rather than API reference pages, so they should be read as vendor descriptions unless separately confirmed in product and security documentation. [3][5]

The public materials name several deployment settings and organizations, but they do not give a full product inventory or a formal packaging matrix. They do indicate use across specialties, care settings, languages, and EHR environments, while leaving exact commercial scope, procurement terms, and per-customer configuration unknown from the public documents. [3][5][7][8]

How the documentation can be read by implementers

For builders, the practical pattern is to choose between a hosted integration path and a server-side API path, then wire encounter context, templates, and authentication into the chosen route. The documentation also shows that some outputs are asynchronous and retrievable through polling or webhook patterns, which matters for workflow design and UI state handling. [2][5][6][8]

Hypothetical demonstration: a health system could send encounter context from its EHR, launch an embedded Nabla session for a visit, retrieve a generated note, and then call note-normalization endpoints to map the final note into structured data. That example is only a workflow sketch; the sources do not promise a particular result quality or implementation outcome.