As the ambient AI market grows more crowded, health systems are demanding more than promising pilots, they need proof.
Becker’s Healthcare spoke with Mac Boyter, revenue cycle management and business solutions research director at KLAS Research, and Josh Margulies, vice president of customer marketing at Suki AI, about what makes ROI evidence credible, what “good evidence” looks like for CIOs and CMOs and how organizations can overcome the operational hurdles of scaling ambient AI enterprisewide.
Editor’s note: Responses have been edited for length and clarity.
Question: Why is third-party ROI validation so important right now, given how crowded the ambient AI market has become?
Mac Boyter:
The reception from clinicians has been very positive. What is just as important is the financial, operational, clinical and patient benefit. What can we draw a causative or correlative link to? That is exactly what this study aimed to answer, looking across three very different organizations: McLeod Health, FMOL Health and Rush University System for Health. It shows what moving from good to great actually looks like and informs future strategy in a meaningful way.
Q: The research surfaced some striking numbers — a 65% reduction in after-hours documentation time and more than $1,000 in monthly revenue lift per provider. What surprised you most?
MB: What stood out to me across all three systems was improved documentation and coding — what we call E&M coding uplift. It was not a top priority for any of these teams going in, but once they saw that downstream impact through the revenue cycle and into billing, it became a direct, measurable connection. That it showed up across three very different organizations is what makes it significant.
Josh Margulies: The fact that McLeod Health can attribute between $1,000 and $2,600 per clinician per month in additional ROI was an unexpected benefit. The system also saw a 5.5% to 6.5% increase in two key patient survey questions: “Do you trust your provider?” and “Did you feel heard?” Leaders said they had never seen that in the history of running those surveys. It shows that a tool built for clinicians can have a significant impact on the patient experience.
Q: For a CIO or CMO evaluating ambient AI vendors today, what does “good evidence” look like?
MB: A lot of teams know they need to evaluate long-term fit, but it is hard to filter that out in the sales process. Look for vendors that can clearly articulate a roadmap beyond the note. Who is coming to you and saying, “We are not just here for today?” Also, make sure your pilot cohort is a healthy mix, tech enthusiasts, early adopters, clinicians at risk of burnout and those who are not. That gives you a fair, holistic picture before committing to scale.
JM: Good evidence is specific, third-party validated and tied to outcomes that matter to your organization. Look for independently verified ROI data, not just vendor-reported numbers. The KLAS study shows those consistent results on documentation time, coding uplift, and revenue per clinician across three very different systems. When the same outcomes show up across different organizations, that’s a real signal.
Beyond the numbers, look for enterprise-scale deployments rather than just successful pilots, and get under the hood on the technology itself. Is the product AI-native, or a workflow tool with AI bolted on? That matters for where it can go over the next two or three years. Transparency is still a differentiator, a vendor who tells you what they do well, what they’re still building, and where they’re investing is a better long-term partner than one claiming to have it all figured out.
Q: For health systems convinced by the ROI case, what are the biggest operational and change management challenges to expect when scaling?
JM: This is not a silver bullet — if clinicians expect a magic solution, they will be disappointed. But if it is framed as saving 30% of documentation time or helping clinicians get home on time, they will build on that. The biggest internal hurdle is often reframing this from a cost center to a revenue driver. Pricing has come down significantly, and whether through more patient visits, better coding, or improved satisfaction scores, there are multiple ways this can pay for itself and become a revenue driver. Getting leadership aligned on that framing is often the difference between a stalled pilot and an enterprisewide rollout.
MB: Arm yourself with data before entering leadership conversations. Whether it is coding uplift, increased patient visits, work RVUs or revenue per clinician, those numbers should be ready for the CMIO, CFO and implementation team. Going in with a clear understanding of what ambient AI can and cannot do, along with validated ROI, helps distinguish a deliberate rollout from one that stalls. The rush to adopt is real, but taking a deliberate approach and choosing partners with validated outcomes is worth the extra time.
Listen to the full conversation on the Becker’s Healthcare Podcast here. To read Suki’s full KLAS ROI report, visit go.suki.ai.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
