Ambient Scribe Assurance

The ambient scribe market has an evidence problem. We built the evidence layer.

Glacis produces independently verifiable proof of what ambient AI scribes do at every patient encounter. For carriers who need to price it. For vendors who need to prove it.

Request the 60-Day Pilot

Fixed fee. Defined scope. Three pilot slots available.

$600M
Ambient scribe revenue, 2025
~1 in 3
US practices that had adopted
ambient scribes by 2025
0
Carriers with per-insured
scribe performance data

Twelve months that redefined the risk.

Ambient scribes are the fastest-adopted clinical AI in history, and they no longer just transcribe. Clinicians increasingly rely on them to guide clinical decisions, yet remain the ones who sign the note and carry the liability for whatever the system gets wrong. The litigation, the insurance exclusions, and the clinical evidence have arrived just as fast.

January 2025
California AB 3030 takes effect, requiring AI disclosure in patient communications. The Doctors Company publishes ambient listening liability analysis in JAMA Network Open.
November 2025
A proposed class action against Sharp HealthCare alleges its ambient scribe recorded encounters without consent, and that the AI-generated notes contained boilerplate language stating patients had consented when, per the complaint, no such consent ever occurred.
January 2026
ISO generative AI exclusions take effect for CGL policies. Chubb, Travelers, and CNA add absolute AI exclusions to standard Tech E&O forms. Silent AI coverage era ends.
February 2026
Epic launches its native AI Charting tool, embedding ambient documentation directly into the dominant acute care EHR.
April 2026
In Washington v. Sutter Health (N.D. Cal., filed April 8, 2026), three patients seeking class-action status allege that Sutter Health and MemorialCare used Abridge's ambient scribe to record and transmit their visits without informed consent, in violation of California's Invasion of Privacy Act, its Confidentiality of Medical Information Act, and the federal Wiretap Act. Abridge, the vendor, is not named as a defendant. The suit follows a similar action against Sharp HealthCare months earlier.
Today
Carriers are insuring millions of AI-scribed encounters with no visibility into what the scribes produce. Vendors are selling into health systems whose legal teams are asking questions no one can answer yet. Both sides need evidence that doesn't exist.

The carrier needs to price it.
The vendor needs to prove it.
The clinician needs to trust it.

For carriers & underwriters

The denominator for ambient scribe risk.

You have loss data. You don't have exposure data. Glacis produces per-insured, per-vendor runtime evidence so you can differentiate risk, design parametric triggers, and price with confidence instead of pricing blind.

  • Hallucination rates by specialty and scribe vendor
  • Consent workflow compliance across jurisdictions
  • Documentation accuracy and drift signals
  • Portfolio-level accumulation visibility across vendors
  • Cryptographically verifiable evidence for claims defense
  • Parametric trigger data for next-generation policy design
For ambient scribe companies

Independent proof your scribe does what you say it does.

Health system procurement teams are asking for evidence you can't produce internally. Glacis gives you third-party attestation that clears security reviews, satisfies legal teams, and differentiates you from every competitor who can't prove the same thing.

  • Third-party attestation for enterprise procurement
  • Evidence of consent execution at every encounter
  • Hallucination detection and accuracy verification
  • Competitive differentiation at the procurement table
  • Better insurance terms through verifiable risk management
  • Defense-ready evidence trail for litigation
For clinicians

Assurance behind every note you sign.

You carry the liability for every note you attest to, and you are relying on a system you cannot personally inspect. Glacis is not another dashboard for you to check. It works in the background, holding the AI behind your documentation to the parameters your organization and the scribe vendor have defined, and producing evidence that consent, oversight, and safety checks are operating as intended at every encounter.

  • Continuous runtime monitoring of the AI behind your documentation
  • Your organization's and vendor's guardrails enforced at every encounter
  • Consent validation built into the ambient workflow
  • Safety measures shown to be operating, not just asserted
  • An independent, verifiable record on file if a note is ever challenged
  • Confidence to rely on the tool you already use

Runtime evidence. Not another audit.

Glacis operates at the inference layer with zero data egress. No patient data leaves the deployment environment. The physician's workflow doesn't change. The scribe works exactly as before. Glacis just produces verifiable proof of what it did.

Step 01

Probe

Continuously test scribe behavior against safety and performance parameters. Hallucination detection. Consent verification. Documentation accuracy. Population drift monitoring as patient mix shifts.

Step 02

Enforce

Enforce the limits the scribe vendor and the health system have already defined. When the scribe drifts outside those declared parameters, Glacis can block or flag the output before it reaches the medical record. Prevention, not forensics. Glacis enforces the policy, it does not set it.

Step 03

Prove

Every attestation receipt is cryptographically signed and tamper-evident. Verifiable by any third party without trusting Glacis, the vendor, or the health system. Evidence that holds up in court, in an audit, and at renewal.

"They come to me with what I call a numerator. They never come to me with a denominator, and it's no good. You cannot price."

Lloyd's underwriter, on why AI liability remains unpriced

What you get.

Exposure Assessment

Hallucination rates, consent compliance, documentation accuracy, and risk stratification across your deployment or insured portfolio. The data that doesn't exist in the market today.

Attestation Receipts

Cryptographically signed, tamper-evident records of what the scribe did at every encounter. Independently verifiable without trusting any party in the chain.

Risk Report

Per-vendor, per-specialty, per-jurisdiction risk stratification. Which configurations are higher risk? Which vendors perform better? Where does consent exposure concentrate?

Defense Package

Timestamped evidence trail ready for claims adjudication, regulatory inquiry, or procurement review. Ends the forensic guessing game.

I use an ambient scribe in my own clinical practice. I've caught it fabricating a medication dosage adjustment in a psychiatric note that I never discussed with my patient. That experience is part of why I built Glacis. The technology works and physicians need it. What's missing is the evidence infrastructure that proves it's working correctly, continuously, at every encounter. That's what we build.

Jennifer Shannon, MD

Co-Founder & CMO, Glacis
Practicing Child & Adolescent Psychiatrist

Three pilot slots. First movers set the standard.

The first carrier with ambient scribe performance data defines the underwriting model. The first vendor with independent attestation wins the procurement review.

Request Pilot Access

jennifer@glacis.io  ·  Fixed fee  ·  60 days  ·  Defined scope