The following was originally published on LinkedIn in January 2025.
By Matthew Ko, DeepScribe CEO and Co-founder
Healthcare technology is reaching a crucial turning point: Ambient AI is breaking through the periphery and into everyday clinical practice. With interest expanding quickly, analysts, technology pundits, and journalists generally agree on two truths about AI-driven documentation:
1. The benefits to providers and patients are significant.
2. Mainstream adoption is an imminent reality.
In an MGMA study, four out of five medical group leaders said they intend to deploy or enhance an ambient AI solution in the coming year.
Behind the scenes, however, there’s a growing tug-of-war as to how the industry perceives ambient AI technology. One camp sees it as a revolutionary tool capable of relieving clinicians from administrative drudgery; the other questions its value or dismisses it as a commodity—cheap transcription software that’s easily replaceable.
As someone who has spent years embedding ambient AI into clinical workflows, I can confidently say that technology alone never “wins” in healthcare; success hinges on truly deep adoption of that technology, integrating it seamlessly into a provider’s daily routines.
Why Commoditization Undermines Real Progress
It makes sense that many organizations are labeling ambient AI as a commoditizing technology, especially when the market of vendors has grown by an order of magnitude this past year, some even offering the service for free (I’ve seen a recent count of more than 90 options). In my view, considering ambient AI a commodity does more than simply devalue its strategic potential—it can actively sabotage it.
When this technology is seen solely through the lens of cost, critical components like customized workflows, specialty-specific models, and at-the-elbow training are often skipped over or absent. As a result, adoption rates remain predictably underwhelming, and ROI never materializes. Even worse, those outcomes then feed the narrative that ambient AI “doesn’t work,” even though the true culprit is inadequate support and integration.
Healthcare is a complex, high-stakes environment. If we reduce ambient AI to a race to the bottom on pricing, we risk overlooking the deeper possibilities it offers—possibilities that depend on thoughtful implementation, clinician buy-in, and a willingness to tackle workflow transformations head-on.
Beyond Documentation: Introducing the Ambient Operating System
If we think about ambient documentation as the technology on-ramp, the real highway is what I like to call the ambient operating system. Once conversations are consistently recorded and transformed into actionable data, you unlock an entire ecosystem of intelligent applications and “agentic” workflows that can reshape care delivery.
To be clear, transcribing patient visits is only the beginning for ambient AI. The real promise emerges when that data propels automations, alerts, and analyses that augment both clinical and administrative decisions—often in real time.
For example:
Revenue Cycle Management (RCM)
Imagine a scenario where an ambient platform not only captures the nuances of a cardiology visit but also instantly cross-references potential coding opportunities for that visit. It flags undercoded procedures, identifies missing documentation for high-complexity encounters, and nudges billing staff about potential reimbursements they might overlook. That means fewer denials, faster payments, and a more confident revenue cycle overall.
Value-Based Care (VBC)
In a risk-based contract setting, every clinical encounter contains data that can affect cost, quality, and patient outcomes. An ambient operating system can parse subtle indicators—such as adherence challenges or social determinants of health—in real time, as the patient speaks. It can give clinicians real-time prompts to arrange follow-up labs or care manager referrals. By intervening earlier and with greater precision, organizations can reduce avoidable hospitalizations and elevate quality scores.
These are just two obvious use cases. Ambient AI also holds the potential to expand natural human capability in the clinical world. For example, it can assist in diagnosing disorders through vocal biomarkers identified during a conversation or provide clinical decision support with access to the entire library of clinical research at the point of care.
The ambient operating system will also naturally manifest differently and serve different purposes based on specialty-specific use cases. In oncology, ambient technologies can accelerate clinical trial recruitment and match patients to life-saving therapies, streamlining access to cutting-edge care.
The leap from note documentation to an operating system capable of automated care orchestration might appear substantial, but the bridge is high clinician adoption. Without it, the system’s most transformative capabilities remain locked away, and organizations risk being left behind—watching from the sidelines as others harness the full power of agentic workflows on the ambient operating system.
This is exactly why we have an obsessive focus on adoption at DeepScribe and are proud to lead the industry with an average 80% clinician adoption.
Adoption as the Value Driver
While many ambient AI tool deployments are accompanied by glowing anecdotal feedback, we’ve learned—both through emerging studies and firsthand experience—that positive stories alone are not reliable indicators of measurable efficiency or productivity gains. The reality is that true value lies in sustained adoption and deep engagement, and these metrics often vary widely.
In my conversations with other health system leaders who have piloted multiple vendors, I’ve heard adoption rates ranging from as low as 20% to as high as 80%. The difference is stark: organizations achieving 70% or greater adoption consistently report stronger returns—more accurate documentation, streamlined workflows, reduced burnout, and happier providers. Two metrics, in particular, serve as the strongest leading indicators of success:
1. Clinical Adoption: The percentage of weekly active users compared to the total number of clinicians for whom the tool was made available.
2. Depth of Engagement: The proportion of patient visits documented using the tool versus the total number of visits available.
It’s critical to note that pilot deployments often see inflated adoption rates because they typically involve clinicians who are already enthusiastic about trying new technology. This can paint a misleadingly optimistic picture. To gain a true signal of how ambient AI will perform enterprise-wide, pilots must include a representative mix of clinicians—those excited to participate and those more cautious about changing workflows.
At DeepScribe, we define a user as having “adopted” when they document at least 50% of their weekly encounters using the platform. These measures aren’t just benchmarks—they’re the foundation of a scalable and transformative ambient operating system.
Despite their importance, clinical adoption and engagement metrics are rarely tracked or published alongside studies measuring the effectiveness of ambient documentation tools. This oversight likely explains the mixed results some organizations report regarding ROI.
In low-adoption environments, positive anecdotal feedback can mask the absence of meaningful, measurable outcomes. But make no mistake: the gap between anecdotal success and measurable ROI will only widen as ambient documentation evolves into the ambient operating system. Organizations that fail to act urgently risk falling irreparably behind (Figure 1).
Adoption is not a passive process. It requires more than simply handing clinicians a tool and hoping for the best. Real adoption demands intentionality: purpose-built integrations within clinical workflows, “white-glove” implementation and support, and flexibility to adapt to the unique preferences of individual specialties and users. Medical groups that resist a check-the-box mentality and instead invest in robust change management and specialized training consistently see ambient AI’s greater rewards.
The reason is clear: deeper, more frequent engagement creates a virtuous cycle. A larger repository of “source-of-truth” data from patient conversations enables a broader deployment of insights at the point of care. This transforms ambient documentation from a utility into a strategic asset—one that enhances care delivery, reduces administrative burden, and positions organizations to lead as the ambient operating system takes center stage.
For those that prioritize adoption now, the future isn’t just promising—it’s transformative. For those that don’t, the opportunity will remain tantalizingly out of reach.
Bridging the Financial Chasm: Aligning Incentives for Innovation
The path from a promising pilot to large-scale transformation is often obstructed by a formidable financial chasm. While ambient documentation pilots frequently generate glowing anecdotal feedback—reducing burnout, alleviating cognitive load, and improving provider satisfaction—they are often accompanied by limited quantifiable ROI. This disparity leads some organizations to question whether the leap to enterprise-wide implementation is worth the investment.
On the vendor side, the market’s fixation on price and the perception of commoditization only deepen the problem. Vendors under pressure to compete on cost alone may strip down their offerings, removing the very elements—like workflow integration, specialty-specific customization, and robust support—that drive meaningful adoption and measurable value. This creates a vicious cycle: low investment leads to tepid adoption, minimal impact, and a lingering perception that ambient AI is just an overhyped expense.
The solution lies in reimagining the business model. Taking inspiration from value-based care, innovative structures such as gain-sharing arrangements or “at-risk” contracts tie vendor success directly to the outcomes they deliver for healthcare organizations. This alignment not only alleviates the up-front financial burden on providers but also motivates vendors to continuously refine their solutions for maximum impact. When both parties have “skin in the game,” the focus shifts from simply deploying technology to achieving tangible clinical and financial improvements.
Importantly, this reimagined model extends beyond ambient documentation. As ambient AI evolves into the ambient operating system, these incentive-aligned approaches become even more critical. The ambient operating system’s potential—powering agentic workflows that automate tasks, support clinical decision-making, and streamline administrative processes—depends on scaling adoption and unlocking its full value across a wide range of use cases. Gain-sharing and similar models create the foundation for sustained innovation, ensuring that both vendors and providers have a vested interest in expanding the system’s capabilities and delivering measurable outcomes.
This alignment isn’t just a financial solution—it’s a strategic imperative. By bridging the gap between anecdotal success and hard ROI, forward-thinking healthcare organizations can fully realize the transformative potential of ambient AI. Those willing to innovate in both deployment and business models will lead the way in evolving ambient AI from a tool into a cornerstone of modern care delivery.
Ambient AI in 2025 and Beyond
This coming year presents a crossroads for care delivery organizations: They can either view ambient AI as a limited, commoditized product or recognize it for the broader, higher-value operating system it can become—one that powers “agentic” workflows and provides real-time insights to support clinical and financial outcomes.
Here’s what forward-looking healthcare leaders should be asking:
- How should we define adoption and depth of engagement at our organization?
- How will this ambient AI solution integrate into our specialties’ workflows and EHR systems?
- What specific nuances and preferences for each service line or specialty should we take into account?
- What training and at-the-elbow support do we need to drive sustainable adoption?
- How can we leverage ambient data for greater value – and which use cases do we want to prioritize on the ambient operating system?
- Which business models best align incentives for rapid, large-scale adoption?
Embracing these questions—and seeking answers—fuels a more strategic view of what ambient AI can offer. By transcending the commodity mindset and partnering with ambient vendors to drive adoption, and even build the ambient operating system, healthcare organizations can unlock the full potential of this technology: evolving from reduced physician burnout and improved documentation quality, to ultimately transforming the way patient care is delivered.
For leaders willing to push beyond superficial deployment and invest in meaningful adoption, the next decade of ambient AI holds extraordinary promise—where every clinical conversation contributes not just to a patient chart, but to a living, intelligent infrastructure that continuously amplifies the reach and impact of modern medicine.
If you’re part of a healthcare organization, what are your experiences using ambient AI? How could it deliver even greater value for you?
text
