Beyond dictation: Why health systems are rethinking the AI scribe

While recent survey data from the American Medical Association (AMA) indicated burnout rates have dropped from pandemic-era peaks, about 42% of physicians still experience at least one symptom of burnout. These symptoms typically include administrative burdens and general job-related stress, and they’re a motivating factor for nearly 45% of physicians who leave their roles, according to a paper in The Permanente Journal. The problem hits primary care especially hard; losing just one PCP has been linked to up to $1 million in lost patient panel revenue and additional physician recruitment costs.

For today’s health systems, investments in clinical technology must be evaluated not just as ways to make the job easier. Tools such as ambient scribes powered by AI need to help organizations reverse these economic realities by supporting staff so they can devote more time to the practice of medicine. The right AI scribe solution should connect with clinical and technical workflows and meet the needs of a diverse set of care professionals.

AI scribes as clinical care partners

A scribe starts with ambient capture of a clinical encounter as it’s happening. Today, voice capture from a clinician’s laptop or tablet in a quiet exam room is table stakes. Leading solutions leverage microphones, supported by on-device data encryption, to enable capture in surgical settings, emergency departments, or other venues where clinicians can’t readily access hardware or establish a network connection.

From there, the scribe generates a structured note for the encounter. Again, modern tools have a leg up on those that simply replicate dictation workflows first defined decades ago.

Driven by AI, scribes can suggest relevant diagnostic codes such as SNOMED and ICD, flag the codes most relevant for billing purposes, and identify any codes that may need to be added or adjusted. Leading scribes also empower clinicians to ask questions during an encounter, drawing on a patient’s record as well as validated guidelines and reviewed research to provide evidence-based answers and document them accordingly.

These features allow the clinical note to evolve from an unstructured narrative to an output ready for the electronic health record (EHR) and other downstream applications. That positions the scribe as an AI care partner and critical layer of clinical decision support, not just a remedy for excessive pajama time.

Customization as the ultimate differentiation

Another limitation of legacy dictation tools often emerges when organizations attempt to deploy them at scale. It’s no secret that different specialties, geographies, care settings, and even individual clinicians have different documentation needs.

General-purpose dictation cannot capture this nuance. It cannot distinguish notes for cardiology, dermatology, endocrinology, and so on, whether it’s the diagnostic codes, structural elements, or data points necessary to provide a complete and consistent record of care. In the absence of nuance, the note is still an unstructured narrative, and often too generic to be of value.

Health systems benefit from scribes built to take advantage of AI and provide customization at scale. Templates designed for specialties and their numerous types of encounters can improve accuracy in documentation, coding suggestions, and alignment with clinical guidelines or billing practices. Organizations gain peace of mind knowing their notes are comprehensive, consistent, and able to stand up to scrutiny, all without asking clinical staff to do more.

Key considerations for AI scribes

Not all AI scribes are created equal. Along with the clinical care features and customization capabilities discussed above, health systems should have a few other considerations as they evaluate their options.

• User experience. A scribe should be intuitive enough that clinicians can use it immediately, without extensive training. It also should be flexible enough to accommodate hundreds of templates, along with users’ unique preferences.
• True cost. Transparent pricing is a must for health systems limiting spending to high-value IT investments. Scribes should also come with clear implementation timelines and straightforward integrations with EHR, practice management, and billing systems that avoid complications and cause costly delays.
• Security and compliance. Medical records are awash with protected health information (PHI). Buyers need assurance that AI scribes store and handle PHI properly, both to ensure compliance with federal and state regulations and to keep patient data out of attackers’ hands.

The most important consideration, though, may be how the product is built. There’s no shortage of standalone AI scribe tools on the market promising rapid gains in productivity and staff satisfaction, along with fast ROI. The claims are impressive, but they may not be enough for health systems that need the functionality, support, and flexibility that only a platform can provide.

About the Author

Nancy, Cibotti, MD, CMIO, Heidi  

Dr. Cibotti is a primary care physician with extensive experience in healthcare leadership and innovation. She received her undergraduate degree from Harvard and medical degree at Brown University. After her primary care residency at New York Hospital she returned to Boston and spent 20 years at Massachusetts General Hospital and on the teaching faculty at Harvard Medical School. Subsequently, Dr Cibotti joined Iora Health where she served as the Market Medical Director and opened 2 new primary care practices serving the needs of frail elders. In the past 8 years, she has held multiple roles at Beth Israel Lahey Health including Chief Innovation Officer of Primary Care and Associate Chief Medical Information Officer. Her accomplishments include launching BILH’s virtual primary care program, implementing ambient AI scribes and helping to lead a system wide Epic implementation. In January 2026, Dr. Cibotti joined Heidi Health as the US based Chief Medical Information Officer.  

About Heidi

Heidi is building an AI Care Partner to expand clinical capacity by supporting every stage of care delivery. In addition to its AI scribe, Heidi has also introduced Evidence, giving clinicians access to trusted medical research at the point of care. Heidi supports more than 2.7 million patient interactions each week in 110 languages from 190 countries. Learn more at heidihealth.com.