If you’ve managed an ambulatory practice, you know immediately where the money leaks: it isn’t just in clinical inefficiency; it’s in the administrative seams that surround every visit. Phone tag, duplicate registration work, eligibility surprises, referral fallout – these are the operational shortcomings that show up as denied claims, missed appointments, and staff turnover. The clinical encounter is just the tip of the iceberg; the real pressure on a practice’s margins emanates from the processes that feed that encounter.

This isn’t a conceptual problem. It’s an engineering problem with measurable inputs, outputs, and, crucially, fixable choke points. Treat the patient journey as a routed system: each touchpoint deliberate, auditable, and accountable.

Three structural interventions produce the clearest, fastest payoff: true pre-visit intake that pushes discrete data into clinical systems, pre-visit insurance verification that eliminates surprises at check-in, and deterministic automation for handling high-volume phone and messaging workflows. Let’s take a closer look at these.

Pre-visit intake as operational infrastructure

Collecting patient data in the waiting room is a relic of an older operating model. The differential value of moving intake upstream is not merely convenience; it changes who does the work and when. When intake is completed before arrival and mapped to discrete EHR fields, it eliminates transcription work, reduces registration errors, and shortens door-to-provider time. More important, it creates structured data that downstream systems — scheduling, billing, quality-measurement — can consume without manual intervention.

From a budgeting perspective, upstream intake reduces variable labor costs at peak times and lowers back-office rework hours that erode margin. From a workflow view, it converts a noisy, interruptive front desk into a predictable queue that can be triaged and batched. The right operational KPI to track: staff minutes freed per completed intake and percent of visits with discrete data populating the chart at the start of the visit.

Real-time eligibility as revenue protection

Nothing eats margin faster than eligibility surprises. A missed copay, a change in benefits, or an unexpected out-of-network tag can turn an otherwise billable visit into a denied claim or an uncollectible balance. Embedding eligibility checks into the pre-visit flow — not as an afterthought at check-in — changes the financial equation. Confirm coverage, present copay expectations to the patient, and capture payment intent before the visit. That sequence reduces downstream denials, improves time-to-cash, and shortens days-in-AR. In multi-site groups, we’ve seen preventable denials fall within a single quarter once eligibility is moved upstream.

Operational leaders should treat eligibility verification as a revenue-cycle control point, not an IT project. Put ownership with revenue operations, instrument denial rates by payer and visit type, and make eligibility success a standing metric in weekly ops reviews.

Deterministic voice automation: complete work, don’t generate more work

Phone volume remains a primary access channel for most patients. The mistake many organizations make is substituting conversational chatbots or generative assistants that “engage” but do not complete their function. The operational gold standard is deterministic automation: rule-based voice workflows that read real slot availability, apply booking rules, and execute transactions – or hand off with full context to a human when needed.

Deterministic automation reduces average handle times, cuts manual callbacks, and keeps the scheduling engine in one place of truth. Automating a broken workflow simply allows it to fail faster. Redesign has to come first. For practices, the practical benefit is operational containment – fewer partial interactions that spawn follow-up tasks and more completed transactions that require no manual closure.

Measure what impacts the margin

Redesign workstreams must be designed for measurable outcomes. Pick 3 to 5 KPIs that link operations to finance, for example: minutes of staff time reclaimed per patient, percentage reduction in no-shows, denial rate by cause, point-of-service collection uplift, and API/data-mapping success rate. Run a 30-90 day pilot with baseline and target thresholds for each metric. Accountability is everything: assign a process owner, instrument changes in the EHR and ACD, and report weekly.

Develop a pragmatic playbook for rollout

Large programs fail when they assume the IT change is the business change. Start small, and own the change management:

• Identify a single pain point (e.g., new-patient scheduling in one clinic).

• Define the financial hypothesis and KPIs.

• Stand up a cross-functional pilot team that includes registration, revenue cycle, clinical leadership, and IT.

• Set governance rules up front: who may change booking rules, how escalations work, and how discrete fields are mapped and validated.

• Train staff on changed workflows and provide a one-page “when to intervene” playbook – so automation becomes a collaborating tool, not a competitor.

The structural, not cosmetic, fix pays dividends

Redesigning the patient journey isn’t a UX exercise; it’s an operating model change. Cosmetic improvements (e.g. sexier patient portals or nicer waiting room signage) might improve NPS in the short term, but they won’t fix leakages. The structural approach – capturing clean data before a visit, verifying eligibility before care is delivered, and automating repeatable phone and messaging tasks with deterministic rules – changes the input mix to the clinical encounter. That, in turn, reduces after-hours EHR work, minimizes denials, and returns staff time to patient care where it belongs.

If you lead operations, your job is to convert friction into flow. Start by instrumenting the work, pick a single pilot that protects revenue, and treat process governance as a first-class concern. Do that, and the clinical encounter stops being an expensive event you must defend and becomes the reason your operational system actually works and is profitable. The organizations that treat front-end workflow as infrastructure, not convenience, will see the difference on the balance sheet.

About Matthew Order

Matthew Order is Vice President of Business Development at Yosi Health, with more than 20 years of healthcare technology and SaaS experience. Previously, he held roles at MEDITECH, athenahealth and Buoy Health, where he focused on marketplace sales, enterprise partnerships, and product strategy and delivery. At Yosi, he leads enterprise adoption across health systems, translating product integrations into measurable operational improvements for practices and patients.

What You Should Know

• The Capital: Ambient Clinical Analytics has closed a $5M strategic funding round. Investors include Mairs & Power Venture Capital and an unnamed Fortune 500 MedTech firm.
• The Leadership Shift: Brian Tufts, a veteran executive with experience at Vantive and Baxter, has been appointed as the new Chief Executive Officer. Founder Al Berning is transitioning to the role of Executive Chairman.
• The Technology: The company’s flagship product is the AWARE platform, an FDA Class II-cleared system built on clinically validated algorithms and technology licensed from the Mayo Clinic.
• The Clinical Focus: The platform delivers real-time clinical analytics and clinical decision support (CDS) across bedside care, command centers, and remote monitoring environments. A major use case is improving protocol adherence for sepsis care, a notoriously difficult and time-sensitive workflow.
• The Goal: The new capital will be utilized to accelerate innovation, expand the platform’s adoption across health systems, and deepen strategic partnerships in high-acuity environments.

What You Should Know

• The Launch: FDB (First Databank) has announced the general availability of FDB MedProof MCP, the healthcare industry’s first Model Context Protocol (MCP) server purpose-built for AI medication decision support.
• The Protocol: MCP is an emerging, open-source standard that defines how AI agents and Large Language Models (LLMs) interact with external knowledge bases. By creating an MCP server, FDB allows health-tech developers to plug generative AI directly into FDB’s continuously updated, clinical-grade drug database.
• The Goal: The platform dramatically reduces integration complexity. Instead of building custom, brittle connections to legacy databases, developers can deploy AI agents that are instantly grounded in trusted medication intelligence, virtually eliminating the risk of the AI “hallucinating” a drug interaction or dosage.
• The Adopters: Early adopters include Artera, a patient communications platform serving over 100 million patients. Artera is using the server to safely expand its AI agent capabilities across major EHRs like Epic, Oracle Health/Cerner, and MEDITECH.
• The Roadmap: FDB is building its own intelligent solutions on top of this foundation. This includes the upcoming FDB Script Agent, which will advance ambient listening by automatically transforming spoken clinical conversations into structured prescriptions within the ambulatory EHR.

What You Should Know

• The Funding: Human performance company WHOOP has raised $575M in a Series G funding round, bringing the company’s valuation to a staggering $10.1 billion.
• The Backers: The round was led by Collaborative Fund and included heavy-hitting institutional participation from Qatar Investment Authority (QIA), Mubadala, Abbott, and the Mayo Clinic, alongside prominent elite athletes like Cristiano Ronaldo, LeBron James, and Rory McIlroy.

The Wearable Designed for Lasting Progress

Moving beyond traditional fitness tracking, WHOOP is positioning itself as a proactive health platform. It leverages more than 24 billion hours of physiological data and purpose-built AI models to predict health risks and prevent chronic disease before it begins.  The health platform includes an FDA-cleared ECG, a Healthspan longevity feature, Blood Pressure Insights, and Advanced Labs blood biomarker analysis. Research shows that people who wear WHOOP daily log more than 90 additional minutes of exercise per week, get over two extra hours of sleep, and have 10% higher heart rate variability.

The wearable company is accelerating rapidly, boasting over 2.5 million members globally. Operationally, WHOOP became cash-flow positive in 2025, experiencing a 103% year-over-year growth in bookings and exiting the year with a $1.1 billion revenue run rate.

Global Expansion

Armed with a fresh $575M, WHOOP plans to hire over 600 new employees to fuel global expansion across Europe, the GCC, Asia, and Latin America. As they deploy this capital, WHOOP isn’t just trying to sell more hardware; they are aggressively trying to scale a new, AI-driven standard for preventative global health.

“Our raise brings together the world’s most sophisticated investors, leading health institutions, and iconic global athletes behind the mission to unlock human performance and healthspan,” said Will Ahmed, Founder and CEO of WHOOP. “We are building the personal health platform that people use to improve their health and livelihood.”

What You Should Know

• The Launch: The Advanced Research Projects Agency for Health (ARPA-H) has announced STOMP (Systematic Targeting Of MicroPlastics), a massive $144 million nationwide program.
• The Ultimate Goal: STOMP aims to develop tools that are fast, affordable, and broadly available, specifically protecting vulnerable populations like pregnant women, children, and highly exposed workers from long-term disease.

Fixing the Data Layer First

While microplastics and nanoplastics (MNPs) have been detected in human lungs, arterial plaques, and brains, the medical community currently lacks precise, standardized tools to measure the burden or fully understand the biological harm of specific plastic types.

Before ARPA-H can fund the removal of these plastics, it has to fix the fundamental data problem. Currently, microplastic measurement techniques produce wildly inconsistent results from lab to lab.

Phase One of STOMP is entirely focused on measurement and mechanism. The goal is to develop gold-standard clinical tests that can accurately quantify a patient’s microplastic burden at scale. Crucially, ARPA-H is bringing in the CDC as an independent validator to ensure the new methods are universally trusted.

This phase will also produce a “risk stratification mechanism.” Not all plastics are created equal; some polymers may pass through the body harmlessly, while others may cross cellular barriers and trigger severe inflammation. By ranking plastics by biological harm, STOMP will finally give the biotech industry a prioritized hit list.

“Microplastics are in every organ we look at—in ourselves and in our children. But we don’t know which ones are harmful or how to remove them,” said Alicia Jackson, Ph.D., ARPA-H Director. “Nobody wants unknown particles accumulating in their body. The field is working in the dark. STOMP is turning on the lights.”  

Phase Two (Removal)

Utilizing the data from Phase One, researchers will draw on pharmaceutical biology and bioremediation science to design interventions that safely and effectively remove accumulated microplastics from human organs.

The agency explicitly states that these removal approaches will draw on “pharmaceutical biology and bioremediation science, run in reverse.” This is a staggering technological leap. Bioremediation is typically used to clean up oil spills or contaminated soil using microorganisms. Adapting those principles to safely bind, neutralize, or extract synthetic nanoplastics from living human brain or lung tissue represents a genuine medical moonshot.

ARPA-H Program Manager Shannon Greene, Ph.D. noted, “It’s physically impossible for us to completely divorce our lives from plastics. They are in everything we touch—our clothes, the materials from which we get our food and water. We need to understand how microplastics are distributed throughout the body and what harm they are causing before we can take the next leap forward to ultimately remove them and improve human health.” 

What You Should Know

• The Baseline Score: In its “First Look” Emerging Insights Report, KLAS Research awarded Abridge Ambient AI for Nursing an overall performance score of 94.3 on a 100-point scale. KLAS notes this is based on limited, early-adopter data from 9 individuals across 6 organizations.
• The Core Functionality: The solution utilizes conversational AI to convert verbalized nursing observations and patient care into drafted flowsheet documentation within the EHR. All surveyed customers validated its integration with Epic.

Designing for Nurses, by Nurses

Nurses are naturally—and correctly—skeptical of AI tools that might hallucinate data into a patient’s chart. To combat this, Abridge prioritized a “Linked sources” feature, which 100% of the surveyed organizations are utilizing. When the AI drafts a flowsheet entry, the nurse can trace that specific data point directly back to the exact moment in the transcript where it was spoken. The AI acts as a highly efficient data-entry clerk, but the nurse remains the ultimate, authoritative editor.

This success also requires a shift in human behavior. Nurses are not used to “narrating” their care out loud. To smooth this transition, Abridge and its early adopters have leaned heavily into change management, with 83% of organizations utilizing mock patient simulation labs prior to go-live. One manager even noted they introduced the new workflow “as a game with some friendly competition” to overcome initial resistance.

“Abridge Ambient AI for Nursing captures the important parts of nursing and patient conversations that need to be filed in discrete places within the EHR and flowsheets,” noted one surveyed VP in the KLAS report. “It essentially moves the documentation burden away and moves documentation to the background so that the foreground is the patient and nurse connection.” 

Expanding the AI Footprint

While a 94.3 KLAS score is an excellent debut, it is important to remember this is an “Emerging Insights” report based on a limited sample size of 6 early-adopter organizations. The true test will be maintaining that score as the platform scales nationally.

Abridge, which secured a massive $300 million Series E round in June 2025, certainly has the capital to force that scale. But the feedback from current users shows exactly where the product needs to go next. 

Future Roadmaps

Customers want the AI to expand beyond just flowsheets to support the entire nursing workflow, including generating comprehensive end-of-shift summaries and documenting patient education.

For more information about the KLAS report, visit https://klasresearch.com/report/abridge-ambient-ai-for-nursing-2026-reducing-nursing-workflow-burden-through-ambient-ai-documentation/3998

What You Should Know

• The Launch: Revenue cycle and clinical AI platform Commure has launched Commure Dictation, an AI-powered “speech-to-cursor” extension.
• Hardware Independence: The tool eliminates the need for expensive, dedicated dictation hardware. Clinicians can simply use the Commure Ambient mobile app as a wireless microphone.

Voice as an Operating System

Instead of toggling between different applications, a physician can now use their smartphone (via the Commure Ambient app) as a universal, wireless microphone. They can dictate a complex clinical note, seamlessly transition to an inbox message to a patient, and then verbally draft a specialist referral without ever changing tools.

More importantly, it features Contextual AI Editing. A doctor doesn’t have to manually click into individual fields to fill out a complex order template; they can simply speak the instructions, and the AI will structure the text and fill the fields automatically.

By owning the “voice” at the very beginning of the patient encounter, they are ensuring that the data entering the EHR is highly structured and optimized for their downstream systems. The text generated by Commure Dictation flows directly into autonomous coding and Clinical Documentation Integrity (CDI) workflows, optimizing revenue cycle operations. 

“Commure Dictation is not just voice-to-text. It is a single AI-native voice platform that works wherever clinicians document,” said Sam Ascher, SVP & GM of Ambient AI at Commure. “For health systems still managing legacy dictation infrastructure, the case for switching and consolidating has never been clearer.”

When ambient listening/generative AI first entered the clinical mainstream, the value proposition was simple: reduce documentation burden and give clinicians their evenings back. That promise resonated. Burnout was rising. After-hours charting had become normalized. Health systems needed relief.

And ambient documentation delivered.

But as deployments scaled from pilot projects to enterprise rollouts, the conversation matured. CFOs and revenue cycle leaders began asking a harder question: beyond time savings, what other sustainable financial and operational returns can AI deliver?

Time savings, while meaningful, were never going to be enough.

Time Savings Is the Starting Point, Not the Endpoint

There is credible evidence that ambient AI reduces documentation time by 20% to 40%, along with large reductions in hours spent charting. Studies have also associated its use with modest increases in weekly RVUs and encounter volume, without increasing denial rates. That stability matters. Productivity gains that trigger denials destroy their own value.

Burnout reduction has financial implications as well. Physician turnover is expensive, costing approximately $4.6 billion each year. Replacing just one physician can cost hundreds of thousands of dollars. That includes recruitment, onboarding, and lost productivity. Even slight improvements in retention protect operating margins.

But documentation efficiency is, at best, an indirect return. Enterprise technology decisions require more than reclaimed time. They require measurable revenue integrity.

That is why the ROI conversation has shifted from note creation to what happens next, from documentation alone to the full encounter-to-cash workflow.

Autonomous Coding is Where ROI Accelerates

Healthcare has lived for years with a quiet revenue drain caused by incomplete documentation and undercoding surrounding missing specificities, uncaptured comorbidities, conservative evaluation / management (E/M) leveling, and risk adjustment gaps.

Ambient AI changes the starting point. It captures richer clinical detail in real time. But the true inflection point occurs when that documentation feeds directly into autonomous E/M and CPT coding, not just suggestions, but structured, defensible coding logic embedded at the point of care within the same unified platform.

That shift has several effects.

First, it reduces underpayment. More complete documentation translates into more accurate leveling and procedure coding. In value-based environments, better capture of hierarchical condition categories improves risk scores in ways that compound financially over time.

In procedural settings and ambulatory surgery centers (ASCs), the impact can be even more pronounced. When AI-generated operative reports feed directly into autonomous CPT coding, procedure capture becomes more precise, modifiers are applied appropriately, and missed billable elements decline. That closes one of the most persistent leakage points in specialty and surgical revenue cycles.

Second, it reduces reliance on retrospective chart reviews and manual coding intervention. Administrative overhead declines as the front end becomes stronger.

Third, coding transparency improves compliance. When logic is embedded upstream, organizations avoid the downstream scramble to reconcile documentation and billing discrepancies.

This is where incremental RVU lift becomes sustainable revenue performance.

The Overlooked Multiplier: Denial Prevention

Even more powerful, and often under-discussed, is denial prevention.

A significant portion of claim denials stem not from clinical appropriateness, but from documentation deficiencies and coding mismatches. Missing medical necessity language. Insufficient time documentation. Procedure-to-diagnosis inconsistencies. Contract-specific nuances overlooked before submission.

If AI stops at documentation, those problems persist. If AI integrates autonomous coding with pre-submission claim integrity screening, denial risk declines before a claim ever leaves the system.

When that screening is contract-aware, incorporating payer-specific policies, authorization requirements, medical necessity edits, and modifier logic, denial prevention shifts from reactive appeals to proactive revenue protection. Claims are validated before submission, not repaired after rejection.

Even marginal reductions in denial rates have a disproportionate financial impact. Rework decreases. Appeals decline. Payment cycles accelerate. Revenue becomes more predictable.

At scale, that stability matters more than marginal increases in visit volume.

Better documentation enables better coding.
Better coding produces cleaner claims.
Cleaner claims reduce denials.
Reduced denials accelerate cash.

When these capabilities operate together inside a unified note-to-bill architecture, the compounding effect becomes structural rather than incidental.

That compounding effect is the real financial story.

Moving Beyond Incremental Gains to Transformational Claims

The industry’s early fascination with ambient AI as a “digital scribe” is giving way to a more strategic perspective. Documentation is not the endpoint. It is the first link in the encounter-to-cash chain.

The organizations realizing meaningful ROI are those integrating:

• Real-time clinical documentation
• AI-generated operative reports in procedural workflows
• Autonomous E/M and CPT coding
• Risk capture optimization
• Proactive denial prevention

When these capabilities operate together, the impact shifts from incremental productivity gains to structural revenue integrity improvement.

That is a different category of value.

The current shift from enthusiasm to financial accountability is healthy. While ambient AI clearly does not double margins overnight, nor eliminate revenue cycle complexity, when embedded within a unified encounter-to-cash platform strategy, where documentation integrity, autonomous coding, and denial prevention operate as one continuous workflow, it does produce an environment where the financial effects become measurable, defensible, and durable.

The ROI question was never really about documentation time savings alone.

It was about whether AI could strengthen the economic foundation of care delivery across the entire note-to-bill continuum.

We are now beginning to see where that answer lives.

About Pat Williams

Pat Williams is a seasoned healthcare executive and the Co-Founder and CEO at iScribeHealth, specializing in healthcare IT solutions. With a rich background in business development, marketing, and operations, Pat has a knack for solving complex business challenges in the healthcare sector. He is passionate about leveraging technology to enhance healthcare delivery and improve patient outcomes.

What You Should Know

• The Regulatory Win: GE HealthCare has received 510(k) clearance from the FDA for True Definition DL, the latest addition to its portfolio of deep learning (DL) image reconstruction solutions for computed tomography (CT).
• The Clinical Target: The technology is specifically designed to drastically improve spatial resolution and clarity in high-contrast regions, making it uniquely suited for lung, musculoskeletal, and inner ear imaging (e.g., finding tiny pulmonary nodules or assessing trabecular bone patterns).

The Sub-Second Chest Scan

With over 80 percent of all health system visits involving an imaging exam, the radiology department is the undisputed diagnostic engine of the modern hospital. But that engine is currently under unprecedented strain.

Driven by an aging global population, the cancer burden is projected to increase by over 75 percent by 2050, and cardiovascular disease is expected to rise by nearly 90 percent. Radiologists are being asked to find smaller anomalies, faster, across a higher volume of patients.

GE HealthCare announced a major technological release to help relieve this pressure, securing FDA 510(k) clearance for True Definition DL. This deep learning image reconstruction tool promises to fundamentally alter the physics of CT scanning, allowing clinicians to generate ultra-high-resolution images of the lungs and bones in under a second.

“Image quality matters in healthcare – because when imaging performance is aligned with the specific diagnostic task, it plays a critical role in improving accuracy, enabling earlier detection, and guiding appropriate patient care,” said Chad Rowland, Executive Director of Global Premium CT and Photon Counting at GE HealthCare.

Defeating the Radiation Trade-Off

If a radiologist needed to get a crystal-clear look at a tiny pulmonary nodule or the auditory ossicles of the inner ear, they traditionally had to make a clinical compromise. Better spatial resolution usually required hitting the patient with a higher dose of radiation, slowing down the scan time, or limiting the physical coverage area of the scan.

True Definition DL uses a dedicated deep neural network to sever this trade-off entirely.

By training the AI to recognize and suppress visual artifacts while simultaneously enhancing spatial resolution across multiple directions, GE HealthCare has created a system that outputs a massive 1024-matrix display. Most impressively, the system is so fast it can capture high-resolution chest imaging in under one second.

As Dr. Stefanie Bitschnau, a radiologist at Radiomed, noted, “In chest imaging, this level of detail is particularly valuable for assessing small airways, supporting earlier and more confident evaluation of interstitial lung disease.”

What You Should Know

• Marathon Health has appointed Chris Pricco as its new Chief Executive Officer. 
• He succeeds co-founder and longtime CEO Dr. Jeff Wells, who will remain with the company as a board member.

Last year alone, Marathon opened 83 new health centers and grew its headcount to 3,300 employees. They are currently expanding their open-access network into heavily populated, highly competitive metro areas like Philadelphia, Detroit, and Baltimore. Navigating that kind of multi-state regulatory, staffing, and real estate complexity requires a very specific operational DNA.

Pricco brings 30 years of heavy-hitting executive experience. He most recently led specialty networks at Paradigm (driving 300% growth over five years) and previously spent over 15 years in senior leadership at Optum, eventually serving as COO of its medical benefit management division.

“Chris understands the complexity of health care and the opportunity ahead for Marathon,” noted Dr. Wells, who will transition to a board role. “He knows what employers need from their benefits investments and is deeply committed to transformative care.”

Primary Care Gets 5% of US Healthcare Funding But Drives 90% of Costs

The underlying thesis of Marathon’s model is cost control through early intervention.  “Primary care accounts for roughly 5% of America’s health care investment, while we spend three to four times that on strictly administrative tasks,” Pricco observed. “Yet primary care sits upstream of – and directly influences – 90% of total care costs.”

That statistic is the entire value proposition of Advanced Primary Care. If an employer can spend slightly more on high-quality, accessible primary care upfront, they can avoid the catastrophic downstream costs of unmanaged chronic diseases and unnecessary emergency room visits.

With a reported 95% client renewal rate, Marathon Health has already proven that its clinical model works and that employers are willing to buy in. The challenge now is execution at a national scale.

What You Should Know

• The Launch: UnitedHealthcare (UHC) has introduced Avery, a generative AI companion integrated directly into the UnitedHealthcare app and myuhc.com.
• The Scale: Avery is currently live for 6.5 million employer-sponsored members and 160,000 Medicare Advantage members. UHC plans an aggressive expansion to reach 20.5 million members across commercial, Medicare, and Medicaid plans by the end of the year.

Beyond the Chatbot: Agentic Healthcare

For years, health insurance chatbots have been little more than glorified FAQ search bars. If a member asked a question more complex than “Where is my ID card?”, the bot would typically fail, forcing the frustrated patient to call a 1-800 number, wait on hold, and explain their entire problem from scratch to a customer service agent.

Avery doesn’t just read your benefits summary back to you. If a member needs to see a doctor, Avery can analyze their specific coverage details, compare care options based on cost, and then—crucially—actually call the network primary care provider to schedule the appointment on the member’s behalf. This transitions the AI from a passive information portal into an active care coordinator.

The operational results of this capability are staggering. UnitedHealthcare reports that 90% of the time, members who engage with Avery are able to resolve their issues without ever needing to speak to a human advocate.

“People want health care to be easier to use and tailored to their personal needs,” said Dan Kueter, CEO for UnitedHealthcare’s commercial business. “Avery is one way we are responding to consumer demand for a more coordinated and simpler experience.”

The Call Center ROI & Handoff

Currently, 90% of the time, members using Avery do not require assistance from a human advocate. When an issue is too complex for the AI, Avery automatically transfers the member to a live customer advocate. But instead of the advocate starting blind, Avery instantly generates a full synopsis of the member’s history, intent, and uploaded documents. The advocate spends less time taking notes and digging through databases, and the member is spared the infuriating task of repeating their problem.

Parent company UnitedHealth Group is investing $1.6 billion into AI this year.

What You Should Know

• The Launch: Clinical AI lab Corti has released Symphony for Medical Coding, an agentic AI model available via API for healthcare software developers.
• Performance Benchmark: The model “claims” to outperform generalized LLMs from tech giants—including OpenAI, Anthropic, Amazon, Oracle, and Google—by more than 25% in clinical accuracy benchmarks.

Beyond the Back Office: The Human Cost of Bad Data

The problem is fundamental: generic LLMs are essentially highly advanced autocomplete engines. They treat medical coding as a simple “prediction” or “labeling” task. But medical coding—which requires navigating over 70,000 diagnosis codes in the ICD-10-CM alone—is not about prediction. It is a highly complex, hierarchical reasoning task that requires strict adherence to constantly evolving clinical guidelines.

When a human coder is working under massive time pressure, they miss the nuance buried in clinical notes. Corti recently ran its system against Danish patient data and discovered it identified three times as many suicide attempts as had actually been coded by humans. The clinical evidence was sitting right there in the medication records and physician notes, but it never made it into the structured data.

When health systems fail to code accurately, they cannot track disease trends, allocate proper resources, or design preventative interventions. “Medical coding has been treated as a back-office cost center for decades. It isn’t – it’s the data layer that healthcare runs on,” noted Andreas Cleve, CEO and co-founder of Corti.

What You Should Know

• The Scale of the Study: In the first major multi-site, multi-vendor study of its kind published in JAMA, researchers analyzed data from 8,581 ambulatory clinicians (including 1,809 AI scribe adopters) across five major academic health systems: Mass General Brigham, Emory Healthcare, UCSF, Yale New Haven Health, and UC Davis. The systems utilized tools from Ambience, Nuance DAX Copilot, and Abridge.
• The Baseline ROI: Across the board, AI scribe adoption was associated with 13.4 fewer minutes of total EHR time and 16.0 fewer minutes of documentation time per 8 scheduled patient hours. It also yielded a modest productivity bump of 0.49 additional weekly visits.

The AI Scribe Reality Check: Why 16 Minutes of Saved Time is Just the Beginning

The new study, co-led by investigators from Mass General Brigham and UCSF Health, tracked ambient documentation use across five U.S hospitals for more than two years. Across over 1,800 adopters, the technology was associated with a reduction of 16 minutes in documentation time and 13.4 minutes in total EHR time per 8-hour shift. The findings are a necessary, sobering reality check for the industry. AI scribes are absolutely working, but they are not magic.

As Dr. Rebecca G. Mishuris, Chief Health Information Officer at Mass General Brigham and senior author of the study, aptly noted, these reductions are “modest” and are unlikely to fully account for the massive drops in physician burnout often associated with these tools.

Other key findings of the study include: 

No Change in “Pajama Time”: Interestingly, the study found that time spent on the EHR outside of scheduled working hours (often referred to as “pajama time”) did not change significantly with the adoption of AI scribes.

The “Power User” Gap: The benefits were highly dependent on utilization. Clinicians who used the AI scribes for 50% or more of their visits saw massive gains, spending 21.3 fewer minutes in total EHR time and 27.3 fewer minutes on documentation. However, only about 32% of adopters actually used the tool that frequently.

Demographic Variations: The reductions in documentation burden were greatest for primary care specialists, advanced practice clinicians, and female clinicians.

The Financial Impact: While the 1.7% increase in visit volume is notable, the associated marginal Evaluation and Management (E/M) revenue generated was minimal, averaging just $167.37 per clinician, per month.

A Reallocation of Time

As the study authors suggest, clinicians may be reallocating those 16 saved minutes back into other critical patient care activities—like reviewing prior documentation, answering patient messages in their inbox, or simply spending more face-to-face time with the patient in the room.

“Ambient documentation use is expanding rapidly across U.S. health care, making it essential to study how these technologies are impacting clinicians in real time,” said lead and corresponding study author Lisa Rotenstein, MD, MBA, an associate professor of medicine at the UCSF School of Medicine, and director of The Center for Physician Experience and Practice Excellence at Brigham and Women’s Hospital. “Our study demonstrates the impact of AI scribes in diverse real-world implementations at multiple sites. It also emphasizes the value of helping clinicians become comfortable with the technology so that they are reaping its full benefits via frequent use.” 

What You Should Know

• The Launch: Ambience Healthcare has released Chart Chat for Nursing, the first EHR-integrated conversational AI tool designed specifically for inpatient nurses.
• The Pilot Partner: Cleveland Clinic, which recently selected Ambience for a massive enterprise-wide AI documentation rollout, is the first health system to pilot this new nursing-specific technology.

The Workflow Solution

Inpatient nurses are the central coordinators of care. Yet, to understand a patient’s current status, they are routinely forced to embark on a dreaded “chart dive”—manually clicking through dozens of disjointed physician progress notes, scattered lab results, and buried orders inside the Electronic Health Record (EHR). Studies show this digital scavenger hunt can consume up to 41% of a nurse’s time in complex care settings.

Instead of spending 15 minutes piecing together a complex patient’s history during shift change, a nurse can simply ask the AI a plain-language question directly within the EHR. Chart Chat  instantly synthesizes the relevant data from across the patient’s entire longitudinal chart.

Preventing AI Hallucinations

Crucially, from a risk-management perspective, Ambience has built strict guardrails to prevent AI hallucinations. Every response generated by Chart Chat is paired with full source citations, allowing the nurse to trace the AI’s answer directly back to the original physician note or lab result. Furthermore, if the AI cannot find the answer in the chart, it is programmed to explicitly admit it does not know, rather than generating a confident-sounding guess.

“Nurses have told us clearly: the problem isn’t just how long it takes to chart, it’s how long it takes to understand the patient in front of them,” said Nikhil Buduma, Co-Founder and CEO of Ambience Healthcare. “Chart Chat for Nursing meets nurses where they already are, inside the EHR, and gives them the full picture of every patient in seconds.”

What You Should Know

• The Local Impact: Oracle is permanently eliminating 539 roles at its Kansas City campus—the former Cerner headquarters. Affected employees were notified in late March, with formal separation dates scheduled between May 26 and June 1.
• The Roles: The cuts heavily impact the operational core of the business, including software developers, system analysts, program managers, IT, and consulting positions. Affected employees do not have “bumping rights” to displace other workers based on seniority.

The Scale of Layoffs

While Oracle has not publicly confirmed the total number of global cuts, internal Slack user counts monitored by employees suggest an overnight headcount drop of 10,000, with some investment bank forecasts predicting the final toll could reach 20,000 to 30,000 workers.

To understand the enterprise impact of these cuts, you have to look beyond the WARN Act filings and into the internal operational reality. According to employee chatter on platforms like Reddit, the dismantling of the former Cerner workforce has been a steady, rolling bleed since 2022.

Oracle has taken on massive debt to finance its AI data center buildouts in a race to secure computing dominance. To balance the books, the company is eliminating the exact mid-level program managers, system analysts, and software developers who keep complex hospital systems running.

“What’s going on with Oracle is similar to what’s going on with a lot of companies in that space,” noted Chris Kuehle, Managing Director of Armada Corporate Intelligence in a statement to Fox4KC. “Almost every job now is going to have to figure out how to deploy AI… or adapt to a sort of a hollowing out of those mid-level jobs.”