VENDOR PROFILE · AI SCRIBES · CLOUD

Ambience Healthcare

An OpenAI-backed ambient documentation and revenue-integrity platform deployed at Cleveland Clinic, UCSF Health, Houston Methodist, and Memorial Hermann. The vendor that won Cleveland Clinic's head-to-head bake-off across 80+ specialties — and the strongest commercial reference point for buyers whose primary pain is coding accuracy and CDI, not just note time.

Health systems
100+

Live customers including Cleveland Clinic, UCSF Health, Houston Methodist, Memorial Hermann, Ardent Health, and Onvida. 200+ supported specialties.

Series C
$243M

Co-led by Oak HC/FT and Andreessen Horowitz in July 2025, with the OpenAI Startup Fund, Kleiner Perkins, and Optum Ventures. Valuation $1.25B.

Coding accuracy
+27%

Relative improvement over 18 board-certified physicians on ICD-10 coding using OpenAI Reinforcement Fine-Tuning. Vendor-run benchmark, May 2025.

KLAS
97.7

Customer satisfaction score in KLAS ambient-AI evaluations, with top marks for product quality and responsiveness.

What Ambience actually does

Ambience markets itself as a "platform" rather than a single ambient scribe — and the framing is accurate. The product is a stack of point-of-care AI surfaces that share the same audio capture, the same chart-aware context, and the same Epic-embedded UI. The components most often quoted in customer deployments:

SURFACE 01
AutoScribe

Ambient note generation across outpatient, ED, and inpatient settings. Specialty- and subspecialty-aware, with templates for 200+ specialties.

SURFACE 02
AutoCDI + ICD-10

Clinical documentation integrity and point-of-care coding. Inpatient CDI launched 2025; the surface that anchors the revenue-cycle pitch.

SURFACE 03
AutoAVS, AutoRefer

After-visit summary in patient-friendly language, plus referral letter drafting from the encounter. Hooks into the same structured note artifact.

SURFACE 04
AutoPrep + Patient Recap

Pre-visit chart summarization and longitudinal "patient recap" — Ambience claims this as the first chart-summarization surface from an ambient AI platform.

The strategic story is integrated revenue cycle: same audio, same model, same EHR sidecar driving documentation, coding, prior auth, and patient communication. That is the surface that justifies an enterprise contract sized to revenue-cycle ROI rather than just clinician hours saved.

Deployment posture

Cloud-only commercial SaaS. The platform is built on OpenAI models, with deep Epic integration as the deployment surface. There is no on-prem tenant and no path to host the inference in a customer-controlled data center.

SURFACE
Inside Epic Hyperdrive + Haiku

Joined Epic Toolbox in 2025, then deepened the integration with native Epic Ambient Module + SMART on FHIR. Clinicians launch Ambience from desktop or mobile without copy/paste.

DATA PATH
OpenAI-powered cloud

Inference runs on OpenAI infrastructure. Available on Microsoft Azure Marketplace for procurement. Encryption in transit and at rest; SOC 2 Type I + II.

EHR REACH
Epic, Oracle Health, athena

End-to-end Cerner / Oracle Health integration was opened to all Cerner customers in 2025. athenahealth supported. Epic remains the deepest surface.

ROLLOUT
Weeks, not months

Once the system relationship is in place, a clinician can be live within weeks. Ardent Health is in an enterprise-wide rollout; Cleveland Clinic announced system-wide rollout in February 2025.

The evidence base

Ambience leans on vendor-run benchmarks and named-system case studies more than peer-reviewed clinical trials. The most-cited numbers a buyer should know:

  • checkICD-10 coding accuracy — Ambience's model, fine-tuned with OpenAI's Reinforcement Fine-Tuning, beat a panel of 18 board-certified physicians on ICD-10 coding by a 27% relative margin on a curated benchmark of complex multi-specialty cases. The dataset was labeled by consensus of four-plus expert clinicians per case. Vendor-published benchmark, May 2025.
  • checkCleveland Clinic head-to-head — Ambience was selected after a pilot evaluating four ambient-AI competitors across 80+ specialties and subspecialties. The system-wide rollout to Cleveland Clinic clinicians was announced February 2025.
  • checkKLAS validation — Customer satisfaction score of 97.7 in ambient-AI evaluations, with high marks for product quality and responsiveness. Ambience also published third-party "ROI validations" tied to coding-accuracy and compliance gains.
  • closePeer-reviewed evidence is thinner than Abridge's. Ambience's case is built on vendor-controlled benchmarks plus customer testimonials. Buyers who need independent peer-reviewed studies on burnout or documentation time should weight that gap when choosing between Ambience and Abridge.

Privacy and compliance posture

Ambience holds SOC 2 Type I and Type II, signs HIPAA Business Associate Agreements, and is listed in the Microsoft Azure Marketplace for buyers who want to procure through their existing Azure spend. Integration with Epic and Oracle Health uses SMART on FHIR and FHIR APIs — modern, standards-based, auditable.

The catch is the same as every cloud ambient scribe: encounter audio leaves the hospital network. The model behind the product is OpenAI's, which means the inference dependency chain runs through a third party regardless of how Ambience itself is governed. For U.S. Epic-using systems with mature cloud security programs and signed BAAs, this is normal. For Canadian provincial settings governed by PHIPA, HIA, or Quebec's Law 25, or U.S. systems hardening their posture beyond the 2026 HIPAA Security Rule floor, an outbound audio path may be the binding constraint that pushes the decision to an on-prem stack.

Strengths and limitations, plainly

STRENGTHS
Where Ambience wins

Best-articulated revenue-cycle pitch in the category — coding and CDI alongside the note. Cleveland Clinic and UCSF are top-shelf reference accounts. Deep Epic Toolbox integration and modern FHIR plumbing. Strong specialty coverage. OpenAI partnership accelerates model iteration.

LIMITATIONS
Where it does not fit

Cloud-only. Inference depends on OpenAI; the customer cannot point the model at their own infrastructure. Peer-reviewed evidence is thinner than Abridge's. The "ICD-10 beats physicians" headline is a vendor benchmark, not an independent study. Enterprise sales motion only — no pilot path for individual clinicians.

Where Ambience fits versus an on-prem alternative

DimensionAmbience (cloud SaaS)On-prem (WalledCare)
Primary value driver Documentation + coding + CDI as one revenue-cycle play. Documentation, discharge, handoff, document Q&A — modular, on customer-owned models.
Model dependency OpenAI inference. Customer cannot swap the model out. Open-weight models (Llama 3.3, Mistral, MedGemma) served via vLLM on customer hardware. Model is replaceable.
Data residency Cloud. Audio and transcripts leave the hospital network. Inside the hospital data center; no outbound API.
EHR integration Best-in-class Epic Toolbox + Hyperdrive/Haiku; end-to-end Cerner; athenahealth. HL7 / FHIR + sidecar; tight for non-Epic EHRs (Oscar, Meditech, custom).
Coding / CDI Lead surface. Most mature offering in the cloud-scribe market. Coding/CDI is buildable but not Ambience-mature out of the box.
Procurement Direct or via Azure Marketplace. Capital purchase + services.

Ambience is the right answer when the buyer is an Epic-using U.S. system whose CFO wants the coding-accuracy story alongside the burnout one. It is the wrong answer when the constraint is "the model and the audio cannot leave our infrastructure."

Where this fits in the WalledCare directory

Ambience is the strongest reference for the integrated revenue-cycle framing of AI Scribes. If your team has converged on "we need scribe + CDI together, and we are on Epic," it should be on the shortlist next to Abridge. If the framing is "we need ambient documentation that runs inside our hospital network," skip the demo and read our on-prem reference stack first.

Quick facts

Founded2020, San Francisco. Co-founders: Mike Ng (CEO), Nikhil Buduma. Backed by the OpenAI Startup Fund from inception.
CategoryAI Scribe + AutoCDI / AutoCoding + AutoAVS + AutoRefer + AutoPrep + Patient Recap.
DeploymentCloud SaaS only, built on OpenAI. SOC 2 Type I + II, HIPAA, BAA. Available on Microsoft Azure Marketplace.
Primary EHREpic (Toolbox member, Hyperdrive + Haiku, Epic Ambient Module). End-to-end Oracle Health (Cerner). athenahealth supported.
Reference customersCleveland Clinic, UCSF Health, Houston Methodist, Memorial Hermann, Ardent Health, Onvida Health.
Coverage200+ specialties. Outpatient, ED, and inpatient settings.
Funding$243M Series C July 2025 at $1.25B (Oak HC/FT, a16z). $70M Series B February 2024. OpenAI Startup Fund, Kleiner Perkins, Optum Ventures.
RecognitionKLAS 97.7 customer-satisfaction score; selected by Cleveland Clinic in 4-vendor head-to-head pilot.
Websiteambiencehealthcare.com

Use Ambience as the benchmark, then decide

If the deciding factor is "scribe plus coding plus CDI inside Epic," Ambience deserves a seat at the table. If the deciding factor is "the audio and the model stay inside our walls," WalledCare's on-prem stack is the alternative built for that constraint.

send Request a WalledCare pilot arrow_back All vendors

Further reading