DeepScribe and Nabla overlap in ambient clinical documentation, but their public materials point buyers toward different starting points. DeepScribe’s published positioning leans heavily toward oncology and specialty workflows, with named EHR pages around Epic, iKnowMed, and OncoEMR plus modules for pre-visit prep, note generation, coding support, and note customization.[1][2] Nabla’s public footprint is split between an EHR-facing product layer and a developer platform, with documentation that emphasizes ambient notes, dictation, coding, EHR write-back, and an API that can transcribe encounters, generate notes, and extract FHIR-normalized data.[8][10] That means this is not a simple feature-matrix contest. For many buyers, the real question is whether they need a specialty workflow product, a general clinical AI layer inside the EHR, or an embeddable API and iframe model for a platform build.[1][10][12]
They are documented at different product layers
DeepScribe’s public pages describe a packaged clinical documentation product tied to specialty care workflows. Its site says the system turns visits into specialty-specific notes, syncs notes into discrete EHR fields through bi-directional integration, and includes pre-charting, coding support, and a customization layer for note style and structure.[2][3] The same materials repeatedly center oncology, while also naming cardiology, gastroenterology, neurology, orthopedics, and urology as additional specialties.[2][3][6]
Nabla’s public materials present two layers at once. On the commercial site, Nabla describes a clinical AI layer that runs inside the EHR and covers ambient documentation, dictation, coding, reading context from the chart, and writing outputs such as notes, vitals, diagnoses, and patient instructions back into EHR fields.[10] In the developer docs, Nabla Core API is described as a toolkit for transcription, note generation, FHIR-normalized data extraction, dictation, dot phrases, and note editing by instruction, aimed at builders such as telehealth platforms, EHRs, and voice-enabled documentation tools.[8][11]
That distinction matters because some organizations are buying software for clinicians, while others are evaluating a component for their own product roadmap. Public DeepScribe pages do not frame the company primarily as an API for third-party builders in the way Nabla Connect and Core API do.[1][12] Public Nabla pages, by contrast, explicitly address EHR vendors and healthtech builders and describe embedding through a single iframe plus API steps for launching sessions and exporting generated output.[10][12]
Workflow scope overlaps, but the visible breadth differs
On documented workflow scope, DeepScribe presents itself as more than ambient note capture. Its pages describe pre-visit intelligence, encounter documentation, bi-directional EHR sync, oncology-oriented coding support across ICD-10, E/M, and HCC, and a customization studio that learns from note edits.[1][2][3][7] One Epic page adds concrete workflow detail: start from a synced Epic schedule, capture the conversation, and review a structured note with coding suggestions and relevant history inside Epic.[4]
Nabla’s public product pages also go beyond ambient note generation, but the shape is different. Its EHR page says clinicians can use documentation, dictation, and coding in the EHR, with chart context flowing into the note and outputs such as patient instructions, diagnoses, vitals, charts, and flowsheet fields written back.[10] Its API docs add another layer of scope that is especially relevant for technical teams: real-time and file transcription, structured note generation from templates such as SOAP and APSO, note editing with natural-language instructions, dictation, dot phrase support, and normalized data extraction into FHIR-compliant structures with ICD-10 and LOINC coding.[8][11]
A buyer comparing the two should avoid assuming perfect symmetry just because both mention coding and EHR fit. DeepScribe’s public narrative is organized around a workflow package for specialty documentation, with pre-chart review and oncology-specific context as recurring themes.[1][2][7] Nabla’s public narrative is organized around an EHR layer plus builder tooling, with a clearer public description of lower-level functions such as note editing instructions, asynchronous endpoints, and normalized data extraction.[8][10][11] Public materials alone do not establish that one vendor lacks functions that the other has not documented in the same way. They do show different levels of technical disclosure.
Practical detail: a useful way to run the comparison
As a suggestion, split the evaluation into two tracks before asking for demos. In the workflow track, ask each vendor to walk through the same visit from chart context through note finalization and coding review, then show exactly what is written back to the EHR and where the clinician edits occur.[3][4][10][12] In the platform track, ask whether your organization is buying end-user software, embedding an AI layer, or both, because Nabla’s public materials clearly support an embedded product route while DeepScribe’s public materials here emphasize packaged specialty workflows.[8][12]
Then write down the questions the websites could not answer. Examples include support by specialty and setting, depth of template mapping in your EHR, what outputs land in discrete fields versus free text, which coding functions depend on product version or region, and how mobile workflows differ from desktop ones.[4][6][9][10] That approach keeps the comparison grounded in documented differences instead of turning missing public detail into assumptions.
Customer fit is where the contrast becomes most practical
DeepScribe’s homepage says it is made for oncology and states that it is used across 90% of community oncology centers, while its integration pages focus on the EHRs driving oncology and list Epic, iKnowMed, and OncoEMR.[1][7] Even where DeepScribe discusses broader specialty care, oncology remains the center of gravity in public positioning, with SmartPrep described as pre-visit intelligence built for oncology care and coding support described in oncology-specific terms on some pages.[2][7]
Nabla’s public positioning is broader by setting. Its EHR page names Epic, Oracle, athenahealth, NextGen, and Greenway, and says the system runs inside the EHR on desktop or mobile.[10] Its API introduction says the product is intended for builders of telehealth platforms, EHRs, or voice-enabled documentation tools, and the Connect page is explicitly written for EHR platforms and healthtech builders that want to launch ambient AI inside their own product.[8][12]
For a provider organization with heavy oncology operations and named interest in Epic, iKnowMed, or OncoEMR workflows, DeepScribe’s public pages may feel more directly aligned to the buying problem being studied.[1][4][6][7] For a health system looking for a general EHR-layer product across multiple EHR environments, or for a software company planning to embed ambient AI into its own application, Nabla’s public materials describe those customer settings more directly.[10][12] That is a positioning difference, not a verdict on deployment success, note quality, or total functionality.
EHR integration claims are concrete, but not equally detailed
DeepScribe publishes named pages for Epic and oncology-focused integrations. Across those pages, it says notes sync to discrete EHR fields, two-way integration keeps data moving, and the Epic deployment can work with Haiku and Canto. A separate page for DeepScribe H.E. says sessions can start directly from Epic Haiku and that notes map to existing Epic templates.[3][4][6] Those are useful workflow details for buyers already standardized on Epic, especially mobile Epic use cases.[4][6]
Nabla’s EHR page names five EHR brands and says the product reads patient history and context from the chart, then writes notes, vitals, diagnoses, and patient instructions back into charts, flowsheets, and clinical fields.[10] For platform builders, Nabla Connect adds a more implementation-oriented model: create an encounter via API, launch Nabla inside an embedded iframe, and export the generated note and patient instructions back into the host EHR.[12]
What public documents do not settle is depth within each named EHR. DeepScribe gives more visible page-level specificity for Epic and oncology-oriented systems.[1][4][6] Nabla gives wider named EHR coverage on its marketing page and more explicit developer workflow documentation for embedding.[10][12] Neither public set here is enough to answer procurement questions like per-field mapping by note type, operational limits by mobile workflow, rollout effort in a specific environment, or which functions are available in every listed integration. Those remain buyer diligence items.
Security, coding, and outcomes claims need careful reading
Both vendors publish trust and security statements, but the wording differs. DeepScribe says its ambient operating system is HIPAA compliant and describes end-to-end AES-256 encryption, de-identified PHI, multi-factor authentication, limited user access, and single sign-on.[1][4][5] Nabla says no audio is stored by default, PHI is never used to train foundation models, retention policies are customizable, and SSO, SAML, SCIM, SMART on FHIR login support, SOC 2 Type 2, ISO 27001, GDPR, and HIPAA are in place.[10][12] These are vendor claims from public pages, not independent security audits contained in this packet.
On coding, both vendors describe support, but the public specifics again differ. DeepScribe says it supports ICD-10, E/M, and HCC coding and pairs that with specialty and oncology language.[2][3][7] Nabla’s EHR page says it suggests E/M, ICD-10, and HCC codes, while the API docs publicly document normalized data extraction into FHIR-compliant structures with ICD-10 and LOINC and note a newer U.S. E/M CPT code generation release in the changelog.[10][9][11]
Both vendors also publish performance or adoption numbers, but buyers should read them as vendor-reported context rather than head-to-head evidence. DeepScribe pages cite metrics such as note approval, clinician adoption, chart closure time, and hours saved daily on selected pages.[3][4][6] Nabla pages cite organization-specific figures and an across-deployments same-day note completion figure.[10][12] Public materials here do not establish comparable study design, population, specialty mix, baseline workflow, or whether the metrics refer to the same product configuration.
What the public record still cannot answer
Several procurement questions remain open after reading these pages. For DeepScribe, public material makes the strongest case around oncology and selected specialty workflows, but it does not fully map how far those workflows extend outside the named specialties, nor does it publicly detail a builder API strategy comparable to Nabla’s Core API and Connect pages.[2][3][12] For Nabla, the public record makes its builder and EHR-layer story clearer, but it does not publicly break out workflow depth inside each named EHR to the same level DeepScribe does for Epic Haiku.[4][6][10]
Public pages also do not resolve note governance questions that matter in practice: who configures templates at scale, how exceptions are handled in subspecialty documentation, what coding outputs are suggestive versus directly mapped into discrete fields, and which capabilities vary by version, geography, or deployment path.[3][9][10][12] Nabla’s changelog makes versioning explicit for API behavior, which is helpful for technical teams, but that still does not substitute for environment-specific validation.[8][9]
A sensible next step is to treat this pair as a segmentation exercise rather than a winner-take-all test. If your team is a provider organization centered on oncology workflow and Epic or oncology-specific EHR paths, DeepScribe’s public documentation is more narrowly tuned to that operating model.[1][4][7] If your team is an EHR vendor, digital health company, or a health system that wants a more general EHR-layer product with a visible API and embedding path, Nabla’s public documentation speaks more directly to that motion.[8][10][12] The unresolved issues are exactly the ones that require demos, architecture review, and customer-reference diligence rather than website comparison.
Sources & references
- EHR Integrations · www.deepscribe.ai
- AI Clinical Documentation Software | Clinical Notes Powered by AI - DeepScribe · www.deepscribe.ai
- AI Clinical Documentation Software | Clinical Notes Powered by AI - DeepScribe · www.deepscribe.ai
- AI Scribe for Epic | DeepScribe · www.deepscribe.ai
- EHR Integrations · deepscribe.ai
- DeepScribe H.E. | Ambient AI directly inside Epic Haiku · www.deepscribe.ai
- DeepScribe AI Medical Scribe | Built for Specialty Care · www.deepscribe.ai
- Introduction | Nabla Core API documentation · docs.nabla.com
- Changelog | Nabla Core API documentation · docs.nabla.com
- Nabla · EHR Integration · www.nabla.com
- Generate normalized data from a note | Nabla Core API documentation · docs.nabla.com
- Nabla Connect — Embed Ambient AI in Your EHR in Days · www.nabla.com