Our Take
Ambient AI scribes work because they stay out of the way: listen, draft, let physicians decide—not replace clinical judgment with automated fact-feeding.
Why it matters
Physician burnout tied directly to documentation load. A measurable 70-minute daily gain and 74% reduction in evening work is proof the problem has a technical solution that preserves human judgment.
Do this week
Healthcare IT leaders: audit your current EHR documentation workflow this week so you can identify which specialties would see the fastest ROI from ambient transcription pilots.
Beth Israel Lahey Health Deployed Ambient AI to Reduce Clinical Documentation Burden
Beth Israel Lahey Health, a Massachusetts system with 14 hospitals, 175+ practices, and 4,700 physicians, implemented Heidi's ambient AI scribe across three cohorts totaling 1,000 clinicians in 47 specialties. The tool listens to patient-clinician conversations in real time and generates structured clinical documentation as a draft for physician review and approval, rather than replacing the clinician's note-writing process entirely.
The rollout was voluntary. The organization provided peer-led training, specialty-specific materials, and over 500 customized workflow templates to support adoption. Because the platform initially ran as a standalone application, clinicians found secondary uses: documenting complex family meetings in rehabilitation settings and handling multilingual patient encounters.
Results were measured after deployment. Clinicians using the technology saved an average of 70 minutes per day on full schedules (company-reported). Seventy-four percent of surveyed users reported significantly less time on documentation after hours. The system now generates over 18,000 documents monthly (company-reported).
Beyond time metrics, 90% of clinicians reported feeling more present during patient visits and making greater eye contact. Eighty-two percent said the tool reduced cognitive load during clinical encounters (company-reported). One clinician described the experience as feeling "liberated from the computer" and "connected to patients" again.
The Real Gain: Presence, Not Just Efficiency
Healthcare systems have wrestled for years with the dual mandate to improve documentation compliance while preserving patient-centered care. Clinicians spend evenings catching up on notes long after the workday ends—a phenomenon called "pajama time" that signals broader burnout risk across the industry. An extra hour of reclaimed clinical time matters because it directly addresses workforce retention and patient satisfaction simultaneously.
The success here turns on a single architectural choice: the AI generates a draft that clinicians must review and approve before it enters the medical record. This preserves physician agency and accountability. It is not full automation; it is assisted documentation. That distinction explains why adoption rates ran high and clinicians reported relief rather than replacement anxiety.
Beth Israel Lahey Health plans to integrate the system with its Epic EHR platform in the next phase, allowing AI-drafted notes to flow directly into the medical record while maintaining physician-centered review. The lesson extends beyond this one deployment: the most effective use of clinical AI may not be replacing human decision-making but creating space for it.
Prepare Your Workflows and Data Governance
If you are a health system considering ambient documentation: audit which specialties carry the highest documentation burden (orthopedics, psychiatry, primary care typically rank high). Identify whether your EHR can surface real-time conversation data securely and whether your malpractice and compliance teams understand the difference between AI-assisted drafting and autonomous documentation.
Run a pilot with volunteer clinicians and measure both time savings and clinical satisfaction. Set clear expectations that the tool generates a draft, not a final note. Build in a review and edit step; do not optimize it away. The adoption win at Beth Israel Lahey came from giving clinicians control, not removing it.
Finally, plan for integration with your EHR before deployment. A standalone application works for adoption but creates friction when notes must be manually migrated into the official record. Bake that handoff into your roadmap from the start.