Author:
The Promise and the Limitation
When Ambient AI rose to prominence in healthcare nearly two years ago1, it was promised as an answer to one of the industry's most deeply rooted pain points: clinician burnout. At the time, nearly half of practicing physicians reported experiencing burnout 2, due in part to the administrative burden of electronic health record (EHR) documentation and the widespread staffing shortages across the industry. The solution was straightforward: technology would listen and transcribe patient interactions in the background, structuring conversations and freeing clinicians from the keyboard.
This kind of automated documentation was compelling precisely because it was focused. A single problem. A single solution. But today, with over 62% of hospitals having adopted ambient AI tools3 and health systems having invested heavily, burnout rates have changed only marginally. Physicians are still leaving their positions 9 years earlier than in past decades4. The stagnation isn't because the technology behind Ambient AI scribes doesn't work—it does. It's because we've been solving for the wrong problem, or more accurately, we've been solving for only one piece of it.
The uncomfortable truth is this: documentation burden is the tip of the iceberg.
While a clinician's day is fragmented by note-writing, it is equally fragmented by prior authorizations that require phone calls and back-and-forth correspondence. By referral letters that must be manually crafted and verified. By overflowing inboxes filled with patient messages that demand response. By order entry workflows that ask physicians to navigate multiple systems and interfaces. By the need to synthesize and review mountains of unstructured narrative content from outside providers. These are not secondary burdens that disappear once documentation is solved. For many clinicians, they are equal or greater contributors to the cognitive load and time burden that defines their day.
Ambient AI, by focusing narrowly on transcription and note generation, solved for a visible problem. But it left the deeper structural burden untouched. That's why adoption metrics look good on a board slide while clinician burnout numbers remain stubbornly high.
The Real Problem: Multifaceted Burden Requires Multifaceted Solutions
"For independent physicians, the promise of ambient AI goes beyond getting their evenings back and spending less time finishing notes after hours," says Elaine Mendoza, CEO of Sevocity, a SaaS Electronic Health Record (EHR) and Practice Management platform built for independent medical practices. "It can also save meaningful time throughout the day. If a physician can spend less time documenting each patient interaction, that could mean having the capacity to see more patients. For smaller practices operating on tight margins, that added capacity can have a meaningful impact on the financial health of the organization. That's a huge opportunity for them."
That observation captures both the promise and the limitation of the current approach. Yes, ambient documentation saves time on note-writing. But if a clinician gains 15 minutes per patient encounter and loses it to manual prior authorization or spends it reviewing and correcting a referral draft, the net burden hasn't actually decreased. The friction has simply moved.
The industry has spent years measuring what AI produces. We should be measuring what it changes. The real measure of progress isn’t how many notes are generated or how often a tool is used. It’s whether we’re meaningfully reducing the work clinicians carry.
The Solution: Ambient Clinical Intelligence as Workflow Redesign
What's becoming clear is that the health systems seeing meaningful results aren't the ones that simply bolted on a documentation tool. They're the ones reconceptualizing the entire administrative burden that surrounds patient care.
That requires moving beyond ambient documentation to Ambient Clinical Intelligence: AI that understands the context of the clinical encounter and puts that intelligence to work across the workflows that contribute to clinician burden. Rather than solving one task in isolation, ACI creates an opportunity to address the work surrounding patient care more holistically, including:
Ambient Documentation reduces the documentation burden by capturing the clinical encounter and helping clinicians generate accurate notes, but documentation is only one part of the work that follows a patient visit.
Assisted Clinical Revenue addresses the administrative and financial workflows that pull clinicians away from patient care, from prior authorization and referrals to coding and other revenue-cycle tasks.These aren't sidebar tasks—for many physicians, prior authorization alone consumes 13-16 hours per week5.
Clinical Operations tackles the operational friction surrounding every encounter, including orders, patient communication, inbox management, and the work required to synthesize information across the care journey.
Clinical Reasoning helps clinicians make sense of the growing volume of information available to them, supporting decision-making and reducing the cognitive burden of synthesizing clinical context.
Together, these capabilities don't just speed up one workflow. They reshape the entire constellation of activities that fragment a clinician's attention and stretch their day beyond the hours they work.
But here's the critical insight: simply building these solutions is not enough.
Measuring What Actually Matters
While traditional adoption metrics show initial momentum, they miss the full picture. Most health systems track adoption—how many clinicians turned the tool on, how many notes it drafted, how many minutes of dictation it replaced. These numbers show up in vendor pitches and board reports, and while they show progress, none answer the question that determines whether a clinician goes home lighter at the end of the day: does this implementation actually remove work from my day?
The risk with a broader Ambient Clinical Intelligence suite is that we replicate the same mistake on a larger scale. We might measure adoption across four pillars instead of one. We might count prior authorizations drafted and referral letters generated and patient messages processed. And we might miss the same critical gap: what happens after the AI generates these outputs, and does the workflow actually lighten the load?
Through Science at Suki, a scientific research initiative focused on ambient clinical intelligence, we are studying these questions with more rigor than the industry has applied so far. Rather than treating adoption as the finish line, the initiative examines how ambient clinical intelligence is actually implemented inside a health system, and what that implementation does to clinician experience, workflow, and the value the system ultimately realizes.
The goal is ambitious but necessary: to understand not just whether clinicians use the technology across these four pillars, but whether using it makes their day meaningfully easier. To measure not adoption, but burden reduction. To ask not "Did we generate an output?" but "Did we remove a bottleneck?"
The Goal
Ultimately, capturing speech from a patient encounter was the easy problem. Reducing the total burden that surrounds and follows that encounter is the harder one, and what actually determines whether burnout numbers move.
The promise of Ambient Clinical Intelligence—that technology can help restore the joy in clinical practice by removing the weight of administrative burden—is real. But realizing it requires that we learn from the gap between Ambient AI's promise and its delivery. We need to build smarter tools across multiple pillars, yes. But we also need to implement them smarter, evaluate them more rigorously, and keep clinicians at the center of that entire process.
The science is just beginning. The opportunity is enormous. And the alternative—continuing to solve only the tip of the iceberg—is insufficient for true impact in healthcare.
References
1 Lukač, D., et al. (2025). "Ambient AI Scribes in Clinical Practice." NEJM Catalyst. The Permanente Medical Group deployed ambient AI documentation to approximately 10,000 physicians beginning in October 2023.
2 American Medical Association (2024). "Changes in Burnout and Satisfaction With Work–Life Integration in Physicians and the General US Working Population." Mayo Clinic Proceedings. Survey of 7,643 physicians found 45.2% reported at least one symptom of burnout in 2023, compared to 62.8% in 2021.
3 Graetz, I., Yang, F., et al. (2026). "Ambient AI Tool Adoption in US Hospitals and Associated Factors." American Journal of Managed Care, 32(1):e25-e30. Among 6,561 U.S. hospitals, 42.4% were Epic users, and 62.6% of those Epic hospitals had adopted ambient AI documentation tools as of June 2025.
4 Chen, S., et al. (2026). "Why Have All the Doctors Gone? Insights Into Early Clinical Departure Among Physicians in the United States: A National Survey." The Permanente Journal. Survey of 971 clinically inactive physicians found the mean age of departure was 48.1 years in 2024, compared to 57.1 years in a comparable cohort studied in 2008—a reduction of 9 years.
5 American Medical Association (2024). "Prior Authorization Survey Results." Survey of 1,000 practicing physicians found physicians and their staff complete an average of 29.1 prior authorization requests per week, taking 14.6 hours to process. For certain specialties, prior authorization can consume 13-16 hours of clinical time per week.
