Jul 14, 2025
6 MIN READ

AI Agent Deployment Insights From Healthcare Leaders (Webinar Recap & Replay)

With all the hype around agentic AI, it can be challenging for business leaders to cut through the noise to determine which solutions to deploy, what use cases to start with, and how to measure success.

Especially for leaders in risk-averse, highly-regulated industries such as healthcare, it’s essential to go beyond the buzz to understand the true impact of AI agents.

To bring clarity amid the confusion and provide real-world insights on deploying healthcare AI agents, leaders at Allina Health, Duke Health, and SoundHound AI joined Healthcare Innovation for a webinar titled, “Healthcare’s 2025 AI Playbook: Turn Agentic Buzz Into Impact” (watch the replay here).

During the webinar, Allina Health’s Chief Digital & Information Officer, Dr. David Ingham, Duke Health’s Chief Innovation & Technology Officer, Dr. Neil Deep Ray, and SoundHound AI’s Healthcare Delivery Director, Nikhil Mislankar, joined forces to share their experiences deploying AI at health systems throughout the US.

Continue reading for 8 key takeaways from the webinar.

Dr. Ingham: I want AI to deliver measurable benefits. In particular, peeling back the administrative burden for our clinicians and our care team members, measurably improving the experience of our patients, and really bringing operational efficiency. Too much technology over the last 20 years has actually added a lot of burden, or facilitated adding a lot of burden, to our patients.

Dr. Ray: The first place we started was to ask, “What is our main pain point?”, and it was in our call center. We just didn’t have the amount of staff for the volume of calls that were coming in, and we knew it was only going to get worse, just because of increasing patients in our system.

Dr. Ingham: We were finding we just couldn’t hire enough people to handle all these calls, and so AI really helped us both improve the experience for our patients and also ease some of the burden on our agents.

Dr. Ray: We focused on the canceling, confirming, rescheduling aspect of appointments. We wanted to take those “low-hanging fruit” use cases off the table so that our call center agents could focus on more of those complex interactions that really do require a human.

Dr. Ingham: We’re very active in the call center space, handling rescheduling appointments, canceling appointments, refills — some of those high-volume, low-complexity phone calls. With our partnership with SoundHound, we actually have about 5% of all our calls now 100% handled by our AI agent.

Nikhil: Having those end-to-end solutions with Epic, or any other EMR — which allows the AI agent to complete the patient’s task for them — really does take your business value and actually make it a business proposition for all the patients.

Dr. Ingham: We’ve had tremendous success with moving appropriate data back and forth and having SoundHound’s system really work hand-in-hand with our Epic system.

Dr. Ray: You can have the best technology in the world, but if it doesn’t integrate with your systems — your phone system, your EMR, etc. — it’s almost useless at that point.

Dr. Ingham: Voice continues to be the primary channel for accomplishing so much within healthcare. We have about 20% of our appointments scheduled online. So, we’re continuing to work there, but voice has got to continue to improve. It is not going away.

Dr. Ray: I think an integrated omnichannel experience is really what people — not just patients — are looking for.

Note: integrated omnichannel is also known as “multimodal,” allowing users to engage with an AI agent using voice, chat, and image modalities all within one interaction to optimize the user experience.

Dr. Ingham: We have a team and process that is still part of our general IT intake process, but if there’s an AI component, a side request is made to the AI team. They take a look at it, and run it through our “responsible use criteria.” Once they’ve checked it through to make sure it fits within our existing platforms, it gets back into our existing IT space.

Dr. Ray: We have a governance board that looks at AI from a safety standpoint and an ethics standpoint, and we also have a separate governance team that looks at it from a use case and business standpoint. We want to see that value that everyone’s talking about before we are trying to implement it.

Dr. Ingham: I advocate for trying to find relatively narrow platforms that can do different things, that have some flexibility but don’t claim, or don’t pretend, to be able to solve every problem across the enterprise.

Dr. Ray: We see a lot of point solutions out there and, not that they can’t be helpful, but sometimes they’re so narrow they don’t actually move the needle in any way. Having multiple platforms that potentially even integrate together, I think that’s what will be key.

Dr. Ray: What I look for is partners that will help us co-develop different use cases in a quick manner, and also be able to scale to new use cases that we haven’t thought about. Secondly, if you have a vendor that has cut their chops in other industries — banking, finance, etc. — there’s a lot of different things that you can learn from those industries.

Dr. Ingham: When I’m looking at any partner, I’m looking for an actual partnership, where certainly I want them to be bringing in a solid platform with solid use cases that will work, but I want them to have some flexibility and be willing to innovate and be a thought partner.

Optimize your patient access operation

AI agents are already improving call center service levels and resolution times for leading health systems. Want to learn how you can join your peers in transforming patient and staff experiences? Check out our Healthcare solution page to learn more.

Interested in Learning More?

Subscribe today to stay informed and get regular updates from SoundHound Inc.

Subscription Form Horizontal