For the Fierce Healthcare editorial team, the HLTH 2025 conference in Las Vegas was a hectic week packed with meetings, interviews and onstage conversations all focused on healthcare innovation. We connected with healthcare executives, founders and clinicians to learn how they are tackling hard problems. We listened to compelling panel discussions and fireside chats targeting issues as versatile as agentic artificial intelligence, GLP-1s, women's health, breakthrough therapies and employer healthcare costs.
Now that we've had some time to sort through our notes, here's a briefing on some of the notable trends we saw at HLTH this year. And, below that, executives share their key takeaways from the conference.
For health systems, AI gets real
There is still a lot of hype around healthcare AI, but conversations have evolved to focus on practical objectives as health system leaders look to the technology to improve efficiency, reduce costs and lighten the load for clinicians and patients.
While healthcare AI companies are quick to develop solutions, health systems are trying to balance their own big AI ambitions with careful deployment.
Cleveland Clinic is working with industry partners to codevelop AI tools and deploy them across a number of operational areas, including early identification of sepsis, more efficient medical coding, matching patients to clinical trials and using AI models to identify repurposable FDA-approved drugs that could be effective against other complex diseases.
"The ability to quickly develop solutions has been pretty impressive, and the companies that we're talking to have a vision to transform healthcare, but, obviously, as one of the top healthcare providers in the world, we have to be very deliberate about how we do that and keeping the patient safe," Todd Schwarzinger, partner at Cleveland Clinic Ventures, told Fierce Healthcare on the sidelines of the HLTH conference. "We're pushing the boundaries in terms of testing these potential applications, but we're going to take our time and be deliberate about deploying it where it clinically impacts our patients. A lot of first use cases are around operations where I think there's less of a risk in making the wrong decision, but, even there, we're very deliberate in making sure that we're safely deploying AI."
At ViVE back in February, the Cleveland Clinic unveiled a five-year partnership with Ambience to deploy its AI ambient scribe technology. "Our deployment of Ambience across the organization was one of the most successful technology launches we've had at the Cleveland Clinic. We've got over 4,000 caregivers using that today," Schwarzinger said at HLTH.
Health system leaders are finding one of the challenges of AI is that solutions can look promising in demos and even in pilot programs, but deploying them into practice and into clinicians' workflows is the true test of whether the tech is a viable solution, Schwarzinger noted.
"We're really intentional about where we pilot and where we spend time with partners. I think we owe that to the companies as well. We want to be a good partner with early-stage companies we're working with as well," he said. "That pilot fatigue is a real thing. I think for companies it can be a real challenge to their success if they spend all their time working through pilots and with providers that just aren't converting into commercial relationships," he noted. "We are testing the boundaries in a lot of respects with these early-stage companies. There are going to be situations where we pilot and when it gets into practice, it just isn't a good fit, but we try to minimize that."
Last year, the Cleveland Clinic appointed Ben Shahshahani, Ph.D., a tech executive with more than 20 years of experience in AI and machine learning, as its first chief AI officer to direct the use of the technology across the health system. Along with working with healthcare AI partners, the health system is looking to build its own AI solutions, particularly for clinical use cases.
"[Shahshahani] put in place a framework and an infrastructure and a team that allows us to build some of these on our own, if we want to, and around the edges where we really want to control, particularly clinically. With the risk there, that's where we'll probably initially build some of that on our own," Schwarzinger said. "I think it's in our best interest, for the long term, to partner with best-of-breed partners. I do think what Ben has brought and the capabilities that we have in-house are making us a better partner and allow us to make those choices around where we pilot."
Tampa General Hospital (TGH), a health system that has been expanding its geographic footprint, has invested heavily in AI to improve patient care and operational efficiency. The hospital is an early adopter of Microsoft's DAX Copilot to streamline clinical documentation.
"People have asked me about the ROI. The expenses around turnover, attrition and recruiting costs—that's where we're seeing it. Just because clinicians are faster at their notes now, we're not adding more visits," Scott Arnold, executive vice president and chief digital and innovation officer at Tampa General Hospital, told Fierce Healthcare at HLTH.
While at HLTH, Arnold said he was evaluating conversational AI solutions among the hundreds of startups on the exhibition floor.
"I think some of these companies out here might be three-year solutions for us before we move to something larger or industrial strength," he noted.
Houston Methodist is making big bets on AI, virtual care and remote monitoring to build the "smart" health system of the future, executives said during an onstage presentation at HLTH. The health system opened Houston Methodist Cypress Hospital this year, equipped with ambient listening and AI tools, along with wider hallways for future robots, touch screens and rooms with on-demand video access to specialists.
"Innovative evolution has never been about technology for its own sake. It's about what the technology makes possible for our patients—a hospital bed that alerts a nurse before a patient is in distress, a small wearable button that continuously monitors vital signs, giving back thousands of hours to our bedside nursing support, a camera system that allows clinicians to connect, collaborate and provide care without interrupting a patient's very important rest, ambient voice that reduces clinician burden and burnout and pajama time documentation," Marc Boom, M.D., president and CEO of Houston Methodist, said during HLTH.
The health system has invested heavily in the past three years to create "smart" hospital spaces with more than 11 million square feet covered by RTLS, Michelle Stansbury, associate chief innovation officer and vice president of IT applications at Houston Methodist, said onstage. "Cameras in all patient rooms, emergency rooms, ORs and expansion into procedural areas, clinics and waiting rooms. The results are clear: new capacity, centralization of scarce provider and staff capacity and better clinical and operational outcomes with fewer capital recruitment requirements.
She added, "The smart spaces have provided not only the infrastructure to allow for our virtual care highway, but they provide the foundation for video AI."
The smart technology is now "part of the furniture," Stansbury said. "The artificial intelligence we gather from RTLS and the ORs have delivered returns many times over their initial investment."
Houston Methodist has built out its virtual connected care capabilities and plans to shift 20% of specialty care to virtual in the next three years.
In the next 24 months, the health system will also move away from clerical data entry to use ambient AI tools for clinical documentation, executives said.
The health system is also using an AI-based "smart concierge" solution to handle appointment scheduling and patient calls, with plans to use the tech for other operational areas. "Hospitals and physician practices are a microcosm of every business infrastructure, from transportation to billing, each is ripe for innovative redesign. The smart concierge is a key arsenal in the business modernization approach," Stansbury said.
"Our future will be filled with drones in the sky, delivering critical medicines, self-service scheduling and payment and AI-based support for patients and staff," she said. "When we talk about business modernizations, it's not just about reducing operating costs and inefficient processes. It's about increasing the quantity and quality of time that our providers have to focus on making patient decisions and delivering the very best care."
Momentum behind agentic AI
AI agents were in the spotlight at HLTH, from voice agents that can automate call centers to agents that tackle administrative tasks and workflows.
"AI is moving so fast now. Agentic AI is one that I'm very bullish on, and I think that it holds tremendous opportunity for us to reduce a lot of administrivia and automate it," TGH's Arnold told Fierce Healthcare.
"I'm spending a lot of time figuring out who's doing agentic and doing it well with great results, and what are those results? We need to move and act faster on these things, but that's completely opposite of how our industry usually runs and operates. All this technology does cost a lot, and some people are very skeptical whether it will really save money," he said.
TGH has applied agentic AI to its call center operations—an area many health systems see as low-hanging fruit.
"The one thing that we're now realizing is we've done a great job governing these things into our environment, but we're now having to figure out how do we monitor it for accuracy and drift. We're reorganizing our technical team to deal with that new challenge," he said. "The other challenge is not underestimating the change management that comes along with agentic AI, because lots of people feel like it's a threat to their job. If we mishandle it, it's going to be misapplied and we're not going to get any benefit, and then nobody wins. We're plowing more calories into how we communicate with our team about it, how we make them a part of it, and then how they begin to see their role, kind of reskilling towards managing AI instead of doing all the heavy lifting."
Houston Methodist has been using agentic AI in scheduling, revenue cycle and hospital follow-up calls, according to Roberta Schwartz, executive vice president and chief innovation officer at Houston Methodist, while speaking onstage.
"In the next three years, we anticipate using AI agents to support most of our key operations. This comprehensive implementation is already in progress. We've done scheduling, registration, consents, revenue cycle, prior authorization and clinical support such as medication refills and answering basic questions," Schwartz said.
The health system is projecting a 25% to 50% reduction in costs associated with these particular business functions, she said.
In some applications, agentic AI can increase costs, Schwartz noted. The health system has seen success with text-based care pathways and won't replace this with AI, but sees opportunities to combine these text-based capabilities with agentic AI.
"What's the key takeaway for agentic AI? The takeaway is that it's new. It can be very expensive if used inappropriately. This is why, when it comes to rolling out agentic AI, Houston Methodist will be very deliberate in how, where and when we will use it. Governing our deliberate use of agentic AI will be a consistent demand that we see a comparatively large ROI when adopting this technology," Schwartz said.
"One of the challenges is that you want to avoid is throwing AI after a broken process, or just picking some small pieces of the process that's already broken, and there are a lot of those in healthcare," Cleveland Clinic Ventures' Schwarzinger said. "What we've tried to do fundamentally is think about, how do we redefine a role or redefine a process around AI? That's where agentic AI can be really impactful."
He added, "Voice agents are ubiquitous. Everybody has it in their demo. It's actually pretty easy to build, but to deploy that into practice in a safe way is really challenging. I think there's a huge opportunity. And it goes back to the capacity situation. One of the biggest things for us is we deal with some of the sickest patients in the world and most complex patients in the world. For us, getting those patients to the right care at the right time is a huge challenge to us, and historically, it's been very manually intensive. Agentic AI can be, from a patient access perspective, a real transformational tool."
Provider organizations also are watching to see what the big health tech players—Epic, GE HealthCare and Philips, to name a few—are doing in the AI space and whether they build out their own agentic AI capabilities, executives said.
Health tech leaders say AI is not yet ready for clinical applications, but they see the promise of it. "When we can bring tools to clinicians at the point of care with patients to help them make the right decisions and get our patients to the right care, I think is going to be really transformational. I think that's going to take time, though. I don't think we're ever going to replace the tangible interaction between patients and caregivers, but I think we can make those interactions tremendously impactful," Schwarzinger said.
He added, "I do think we're going to have agentic support for patients over time. You're going to have something in your pocket that will know your entire medical history and will be able to respond to questions that you have with the wealth of all the knowledge of what you've been through from your healthcare journey, and really help you make the next best choice."
Big tech players building applications for healthcare
The big tech companies have long been providing the tech infrastructure backbone for healthcare, but they are increasingly moving into the application layer.
TGH extended its long-term partnership with Palantir to use its Artificial Intelligence Platform to support a Care Coordination Operating System. Since its initial partnership in 2021, the health system has expanded its use of Palantir’s software from one to more than a dozen use cases across the health system, according to a 2024 press release.
Cleveland Clinic has also been working with Palantir for a number of years. The health system developed a virtual command center to optimize patient flow, staffing and resource allocation, which was built on the Palantir's Foundry AI platform. The use of AI has enabled Cleveland Clinic to reduce unused OR time by 40%, and through its virtual command center, Cleveland Clinic has transferred about 30 more patients to the main campus each week, a 7% increase.
HCA Healthcare collaborated with Google Cloud to develop an AI-powered, nurse-designed handoff tool aimed at ensuring accurate, consistent communication.
Amazon Web Services recently rolled out two services for building and leveraging agentic AI on the AWS platform. AgentCore is an enterprise-grade platform and infrastructure service for developers to build, deploy and operate highly capable AI agents securely and at scale, the company said.
Healthcare AI company Innovaccer developed its agentic healthcare intelligence platform, Innovaccer Gravity, using Amazon Bedrock AgentCore.
AWS Quick Suite is a unified, agentic-AI-powered digital workspace designed for business users to interact with company data and applications using natural language.
"AWS Quick Suite actually works in a customer's own environment," Rowland Illing, Ph.D., global chief medical officer and director, healthcare and life sciences, told Fierce Healthcare at HLTH. "What we wanted to do is build an end-to-end agent workflow engine. You can import data from all of these different sources in a secure environment, and then enable end users basically to build agents on the fly."
"I want to see agents in the wild and I actually want to see the impact that agentic AI has on customer outcomes and patient outcomes. That's what I'm not seeing at the moment, but I think we're going to see that next year," Illing added.
Healthcare and life sciences are key industries for AWS, he noted. "We've been making a lot of investments into healthcare and life sciences. You can see that through things like the tech investments we've made at a service level around data types specific to healthcare," he said, noting services such as HealthLake, HealthScribe and HealthOmics.
"None of those are end-to-end solutions, but they are core components of healthcare technology solutions in the future," Illing said.
AWS also developed its own AI-native contact center, and Jupiter Medical Center in Florida is using the technology to replace its legacy contact center. As a result, the center cut a persistent backlog of 1,800 radiology appointments by 60%, Illing said.
The hospital also reduced the rate of abandoned calls—patients who hang up before reaching an agent—from more than 9% to less than 4% through automation, he noted.
Executive takeaways
"At this year’s HLTH, one theme stood out: Real progress in healthcare isn’t just about adopting new technology—it’s about executing with discipline," said Greg Miller, vice president of business development and marketing at Carta Healthcare. "Across the board, leaders are shifting focus toward tools that deliver tangible results. They’re looking for ways to streamline workflows, reduce or eliminate manual effort, and turn clinical data into insights that drive both financial and operational impact. The conversation has moved beyond innovation for innovation’s sake. What matters now is whether a solution can improve performance and help organizations stay resilient for the long haul."
"The next wave of health innovation winners won’t be defined by funding rounds or marketing—they’ll be defined by speed to evidence. HLTH underscored that success now depends on proving outcomes in the real world, not just promising them. The companies integrating AI, real-world data, and patient-centric design into their clinical research are setting a new standard for credibility and speed," said Malcolm Fogarty, a strategic advisor to Lindus Health.
"This year, I heard a lot of fatigue around endless pilots that don’t deliver real outcomes. There was a clear consensus that the next phase of AI in healthcare has to focus on accountability and ROI. We need models that can be tested, tracked, and audited in real time. If you can’t measure impact, it’s not innovation. The organizations that produce ROI in the first few months of implementation will set the new benchmark for what successful AI looks like," said Fawad Butt, CEO and co-founder of Penguin Ai, a healthcare AI company.
"Healthcare’s financial challenges demand more than incremental change. While many organizations are advancing digital enhancements to streamline workflows, inefficiencies in revenue cycle management (RCM) continue to erode margins and drain resources. Artificial intelligence—including GenAI and agentic AI—represents a true breakthrough, transforming RCM into an intelligent, adaptive system that drives efficiency, accuracy, and long-term sustainability. Embracing AI is no longer optional — it’s a strategic imperative for the future of healthcare," said Gautam Char, chief strategy officer at Omega Healthcare.
“What stood out to me at HLTH was the convergence of science, AI and market urgency. We are witnessing the emergence of new therapies, digital biomarkers and scalable platforms all advancing simultaneously, which means the choices we make now will shape care for the next decade," said David Bates, CEO and co-founder at Linus Health, an AI-driven brain health company.
"What I took from HLTH 2025 is a sense of momentum and maturation in the digital health space," noted Michael Dalton, CEO and founder at Ovatient. "Three years ago at HLTH, there was a lot of hype. We’re now entering a stage where conversations about topics like GLP-1s, weight management, and women’s health have evolved toward proven solutions that are informed by the evidence we have seen play out. I've seen this firsthand, where almost 70% of our virtual primary care patients are women, and nearly half are navigating their weight management journeys with us."
He added, "This evolution and success is not a fad. It's a signal of what patients want: trusted, coordinated, whole-person care that offers a seamless connection between virtual and in-person care teams."
“What stood out to me was the sense that healthcare leaders finally see AI as part of the system, not something on the side. Organizations are combining human expertise with data-driven technology that is scalable, cost-efficient, and accurate. The focus has shifted from potential to performance, and that’s real progress," said Kim Perry, chief growth officer at medical AI company emtelligent.
"This year’s discussions reflected a shift in how health systems see value in technology: Innovation is being measured by operational impact, not novelty. The most forward-looking organizations are treating data and AI as operational assets, using them to improve consistency, anticipate patient and clinical needs, and accelerate diagnoses and downstream care. It’s less about what’s possible and more about what’s working at scale," said Conrad Gudmundson, vice president of strategy at Lucem Health.
"We saw a strong focus on how AI is being applied in practical, meaningful ways across healthcare. These conversations highlighted how AI can ease the burden on clinicians by reducing unnecessary noise, while also helping patients better understand and manage their therapies. The future is about AI not replacing clinicians, but supporting them with tools that make care safer and more personal," said Virginia Halsey, senior vice president of strategy and product management at FDB (First Databank Inc.).
"We took part in many conversations on the growing need for AI-ready data in healthcare at HLTH 2025. Delivering raw information alone isn’t enough—it must be organized, contextualized and actionable at the point of care. When data is intelligently structured, clinicians and developers can focus on what matters most, driving faster decisions, better coordination and healthier outcomes," noted Bob Watson, CEO of Health Gorilla, a designated QHIN under TEFCA.
"At HLTH, it was clear that healthcare is evolving toward continuous intelligence powered by AI and advanced analytics woven into everyday workflows. The ability to access accurate, real-time data within EHRs is empowering clinicians to make decisions that are both faster and more precise. Hospitals are leveraging this rich, reliable data not merely for compliance or reporting, but as an essential instrument for improving care quality and transforming performance at the point of delivery," said Kem Graham, vice president of growth and strategy at CliniComp.