The agentic AI conversation at HIMSS 2026 was not confined to the EHR vendors — it was everywhere. In revenue cycle management, payer operations, consumer health, clinical research, and patient-facing access, autonomous AI systems capable of executing multi-step workflows without human intervention at each stage moved from conference-floor talking points to documented operational deployments. The breadth was striking: from a Google Cloud ecosystem producing $27.9 million in measurable enterprise value to a partnership integrating ambient AI documentation directly into revenue cycle management, to Microsoft Copilot quietly handling 50 million health questions daily. The shift was measurable, and it was cross-category.
HotSpot Take
HIMSS26 didn’t just debate agentic AI — it quantified it. From $15 billion in prevented claim denials to 50 million daily health questions handled by Microsoft Copilot, the receipts arrived across revenue cycle, consumer health, clinical intelligence, and patient access simultaneously.
Google Cloud: Five Partnerships, One Strategic Signal
Agentic AI deployments at HIMSS26 moved from pilot programs to production scale, with vendors presenting outcome data from named health systems and payers. Photo courtesy of HIMSS26.
Google Cloud made one of the most consequential HIMSS26 appearances — not through a single announcement, but through a coordinated portfolio of enterprise partnerships that collectively demonstrated the breadth of Gemini-powered agentic AI deployments across the healthcare value chain. Five major organizations announced active commitments: Humana, CVS Health, Highmark Health, Waystar, and Quest Diagnostics, spanning payer operations, consumer health, revenue cycle, and diagnostic services.
“The next chapter is bringing multi-agent support to employees.” — Richard Clarke, Chief Data and Analytics Officer, Highmark Health
The outcomes across those partnerships were specific and substantial. Highmark Health’s internal generative AI assistant, Sidekick, scaled from 1 million to more than 6 million prompts across 74 active use cases in just over a year, delivering an estimated $27.9 million in AI-enabled value in 2025, according to the company. Highmark’s chief data and analytics officer, Richard Clarke, described the direction at the conference: “The next chapter is bringing multi-agent support to employees.”
CVS Health announced Health100, a new health technology subsidiary built on Google Cloud’s Gemini models, positioned as an AI-native consumer engagement platform designed to connect patients across pharmacies, care providers, insurers, and digital health solutions regardless of which vendors they use. Humana’s Agent Assist, supporting more than 20,000 member associates who handle up to 80 million calls annually, and Quest Diagnostics’ AI Companion — which helps patients interpret lab results in plain language through the MyQuest app — rounded out a portfolio that spans patient engagement, workforce support, and diagnostic access.
“It is an agentic era for healthcare, and the shift in the technology is not what AI can do, but what AI is trusted to do.” — Aashima Gupta, Global Director for Healthcare Solutions, Google Cloud
Aashima Gupta, global director for healthcare solutions at Google Cloud, framed the defining tension at the heart of the category: “It is an agentic era for healthcare, and the shift in the technology is not what AI can do, but what AI is trusted to do.”
Voice AI Agents: SoundHound and the Patient Access Layer
The agentic AI story at HIMSS26 extended well beyond revenue cycle and clinical documentation into the patient-facing contact center — the layer of the healthcare enterprise that absorbs enormous call volume before a patient ever interacts with a clinician. HealthTech HotSpot covered this category in depth through our interview with SoundHound AI and MUSC Health at the conference. SoundHound’s Amelia platform, deployed as “Emily” across MUSC Health’s 16-hospital system in South Carolina, has handled more than 2.2 million calls with a 25 percent full automation rate and a patient satisfaction score of 4.4 out of 5 — handling patient access, revenue cycle, and pharmacy inquiries autonomously, at scale, around the clock. HealthTech HotSpot’s full coverage of SoundHound at HIMSS26 details how MUSC Health turned a staffing shortage into a voice AI deployment that now serves all 46 South Carolina counties.
RingCentral entered the same category from a different angle with the launch of AIR Pro for Healthcare, a voice-first omnichannel AI agent platform with over 80 EHR integrations. According to the company, AIR Pro acts as an autonomous digital front door — handling inbound calls, verifying insurance, and scheduling appointments without live staff intervention. The RingCentral launch, alongside SoundHound’s MUSC deployment, signals that the patient access contact center has become one of the most competitive battlegrounds for agentic AI in healthcare.
Clinical-to-Financial Integration: Heidi’s R1 Partnership
A related thread — connecting clinical AI documentation directly to revenue cycle automation — was at the center of Heidi’s HIMSS26 announcement. The ambient AI scribe company announced a strategic partnership with R1 at the conference, integrating its clinical documentation workflow into R1’s Phare Revenue Operating System. The result gives clinicians real-time visibility into payer policy, prior authorization rules, and insurance eligibility at the point of care, while optimizing notes for accurate coding and billing. As HealthTech HotSpot reported, the partnership addresses what both companies describe as a fragmented care-to-cash workflow that costs the healthcare system billions annually — a claim attributed to Heidi CEO Dr. Thomas Kelly. Both the SoundHound and Heidi stories illustrate a broader HIMSS26 pattern: agentic AI is attacking the administrative surface of healthcare from multiple angles simultaneously, not just from within the EHR.
Waystar and the Autonomous Revenue Cycle
The revenue cycle was the most data-rich agentic AI category at HIMSS26, and Waystar arrived with the largest single number on the floor. The company, which expanded its partnership with Google Cloud to accelerate its autonomous revenue cycle capabilities, reported that its AltitudeAI platform has helped providers prevent more than $15 billion in denied claims and reduced time spent on denial appeals and documentation workflows by 90 percent.
“Waystar connects over one million providers to every major payer, powered by more than 100,000 live integrations across electronic medical record systems. That scale means nearly 60% of the US patient population flows through our platform each year.” — Chris Schremser, CTO, Waystar
Chris Schremser, Waystar’s chief technology officer, put the scale in context: “Waystar connects over one million providers to every major payer, powered by more than 100,000 live integrations across electronic medical record systems. That scale means nearly 60 percent of the US patient population flows through our platform each year.” Administrative waste in U.S. healthcare, Schremser noted at the conference, runs to approximately $440 billion annually — framing the revenue cycle automation opportunity as one of healthcare’s largest addressable problems.
FinThrive, XiFin, and Innovaccer: Agents Attacking the Back Office
Three additional vendors made substantive agentic AI announcements at HIMSS26 that illustrate how broadly the autonomous workflow category is expanding across the claims and denial management process.
FinThrive positioned agentic AI not as a product feature but as an operating model, built on its Fusion data architecture. The platform enables autonomous identification of financial risk and execution of workflows across more than 50 revenue cycle use cases. According to the company, early adopters have recovered approximately 1.1 percent of underpayments — representing close to $1 million in recovered revenue within three months of deployment.
XiFin debuted its Empower AI ecosystem, centered on an autonomous Appeals Agent designed to handle the full denial management workflow end to end: reviewing denied claims, retrieving medical necessity documentation, drafting patient-specific appeal letters, and submitting them to payers — all within defined compliance guardrails and without requiring human intervention at each step.
Innovaccer’s Flow Capture addressed a related gap: combining ambient documentation with automated medical coding to handle approximately 80 percent of patient encounter coding autonomously, routing complex cases to certified coders — closing the loop between clinical documentation and revenue capture at the point of care.
The Consumer and Patient-Facing Frontier
Beyond enterprise health systems, several HIMSS26 announcements targeted the consumer-facing layer of the healthcare system — the point at which patients first encounter the healthcare enterprise, often long before they interact with a clinician.
Amazon’s launch of Health AI for Prime members was the highest-profile consumer agentic AI move at the conference. Built on Amazon Bedrock and connected to the nationwide Health Information Exchange, the agent provides personalized triage based on a patient’s longitudinal medical history. By offering eligible U.S. Prime members up to five free direct-message consultations, Amazon is extending agentic AI into consumer-facing primary care access at a scale that no health system can match. The announcement landed differently than a typical vendor launch: Amazon’s existing relationship with hundreds of millions of consumers, combined with HIE-based health data access, creates a patient engagement surface that is structurally new.
Samsung and b.well Connected Health made a significant interoperability-meets-consumer-AI announcement at HIMSS26, framed around the CMS “Kill the Clipboard” initiative. Samsung Health users can now securely download their longitudinal health records to Galaxy smartphones using FHIR standards and CLEAR identity verification. b.well’s conversational AI, bailey™, translates the clinical data into plain language, allowing patients to share their records directly with hospital EHRs — eliminating repetitive intake forms at the point of care. The combination of a major consumer device manufacturer, an interoperability platform, and a conversational AI layer represents a new architecture for patient data portability.
Microsoft provided another consumer-facing data point through a report released at the conference analyzing 500,000 Copilot health conversations. The findings showed Copilot is handling 50 million health questions daily, with queries spiking late at night when clinics are closed — a pattern the report described as the “midnight triage nurse” phenomenon. One in seven searches is conducted by the “sandwich generation” managing care for both children and aging parents. The data, while not a product announcement, offers a real-world signal about how large-scale consumer AI is already reshaping when and how people seek health information.
Clinical Intelligence at the Point of Care: Atropos and Microsoft Dragon Copilot
One of the more clinically significant announcements at HIMSS26 came from Atropos Health, which integrated its Evidence Agent with Microsoft Dragon Copilot in a live deployment at Stanford Medicine. The system listens ambiently to patient visits and proactively generates personalized, evidence-based clinical summaries from real-world patient data directly within the EHR — providing clinicians with relevant evidence at the moment a clinical decision is being made rather than requiring a separate lookup.
The Stanford deployment is noteworthy for two reasons. First, it is a named, live deployment at a major academic medical center, providing independent validation that the integration works in a high-complexity clinical environment. Second, it represents a new model for clinical decision support: rather than a physician initiating a search, an agentic system surfaces relevant evidence based on what it hears during the encounter. Wolters Kluwer reinforced the same theme with its integration of UpToDate Expert AI directly into Microsoft Dragon Copilot and Microsoft Teams, embedding expert-curated clinical intelligence into the workflows clinicians already use as a guardrail against AI-generated errors.
The Multi-Agent Evidence Base: Mount Sinai Research
Academic medicine added independent validation to the agentic AI conversation at HIMSS26 through a study from Mount Sinai Health System, published in npj Health Systems and presented during the conference. The research found that distributing clinical tasks among multiple specialized AI agents is up to 65 times more computationally efficient than relying on a single generalized model under heavy workloads, while also delivering meaningfully higher accuracy.
The finding matters for health system procurement decisions because it provides peer-reviewed evidence for the multi-agent architectural approach that vendors, including Google Cloud and Epic, are building toward. For health system IT leaders evaluating whether to deploy single general-purpose AI models or invest in multi-agent architectures, the Mount Sinai data offers the kind of independent validation that vendor-reported outcomes cannot.
AI Governance: A New Product Category
Perhaps the most structurally significant signal at HIMSS26 was not any individual agentic AI announcement, but the emergence of AI governance as a standalone vendor category. When autonomous AI agents can access patient health records, submit prior authorization requests, draft clinical documentation, and initiate denial appeals — all without human sign-off at each step — the compliance and liability implications require systematic oversight that existing IT governance tools were not designed to address.
Singulr AI launched Agent Pulse specifically to fill that gap: a real-time runtime governance platform providing context discovery, risk intelligence, and policy enforcement to ensure AI agents execute only authorized actions within defined parameters. The product addresses a new category of enterprise risk that has no direct analog in earlier healthcare IT governance frameworks — one where the entity requiring oversight is not a human user but an autonomous software agent operating on protected health information.
Healthcare Dive’s post-conference analysis captured the underlying concern from cybersecurity experts: as AI agents proliferate, managing non-human identities and their system access becomes a new category of enterprise risk that health systems are only beginning to confront. The emergence of a dedicated governance vendor at HIMSS26 is a leading indicator that this risk is being taken seriously enough to warrant commercial infrastructure.
What the Numbers Mean
The agentic AI story at HIMSS26 is ultimately a story about scale and trust arriving simultaneously. The outcomes data — $15 billion in prevented denials, $27.9 million in measured AI value, 90 percent reduction in denial appeal time, 50 million daily health questions handled by Microsoft Copilot — comes from organizations operating at genuine enterprise scale, not from controlled pilots. That matters because the central objection to agentic AI adoption in healthcare has not been capability skepticism but deployment skepticism: can these systems perform reliably at production scale, in regulated environments, with real patient data?
The HIMSS26 answer, from vendors and health systems alike, was increasingly yes. The follow-on question — whether those systems can be trusted to act autonomously within appropriate governance guardrails — is what Agent Pulse, MCP servers, and the broader AI governance conversation are designed to address. As HealthTech HotSpot covered in our HIMSS26 EHR roundup, the EHR vendors are building the same trust infrastructure from the inside of the clinical workflow. The agentic AI vendors showcased this week are building it from the outside in. And as our coverage from HLTH 2025 last October showed, this moment has been building for some time — what changed at HIMSS26 is that the industry stopped debating whether agentic AI works and started presenting the receipts.
Photo credit: HIMSS26
— This original article was created with AI support.
