Last year, I interviewed executives about the results of a survey of CIOs and CMIOs about their AI strategic directions. This year AI-driven care operations company Qventus has conducted a second survey of more than 60 CIOs, Chief AI Officers, and CMIOs, and I spoke with Matt Anderson, M.D., M.B.A., CMIO at HonorHealth Medical Group in Arizona, about his perspective on AI strategy and vendor relationships. 

The Qventus survey found that 94% of leaders say delays in operationalizing AI would put their organization at a competitive disadvantage — and 77% say even a one- to two-year delay would mean meaningful lost savings and efficiency gains.

Anderson discussed the approach to choosing use cases and vendors at HonorHealth, an integrated network in Arizona with nine hospitals and hundreds of primary, specialty and urgent care clinics in the Phoenix area. The health system has one instance of Epic across the board, and Anderson focuses his efforts mostly on the ambulatory and the virtual care spaces. 

Healthcare Innovation: One of the focal points of this Qventus survey is relationships with AI vendors. The survey says that 74% of respondents cite reliance on the EHR vendor’s AI roadmap as an execution obstacle. In addition, in 2025, 52% said they would wait for an EHR feature, while in 2026 only 22% said they would — with 40% ready to deploy a proven third-party solution now. Does that match up with your experience? Do you have to weigh the pros and cons of waiting for Epic to do something vs. working with a startup?

Anderson: That is a constant discussion, because the EHR in general is a pretty sunk cost. You've already paid for it. Some of the AI bells and whistles are maybe marginal increases in cost, but the big cost is already there. So there is some incentive to focus on the EHR as our developer for AI tools, but they can't do everything. 

When you have a problem and you try to find a solution to that problem, you're finding the best tool for that problem. That could be our EMR; that could not be our EMR. For instance, that conversation comes up a lot with ambient. People have been talking about these ambient tools built into the EHR for a while, and now they have some. We were not 100% sure that that was the right tool for us, so we went with an outside vendor that provides ambient when it comes to some of the throughput work that we're doing in in the hospital. That may not be what our EHR vendor is best at.

HCI: If you decide to go with other vendors' tools, then you're managing multiple vendors. The Qventus survey report said that 25% of respondents were managing four to seven AI vendors. What are some of the challenges with that?

Anderson: I think we're used to managing lots of vendors anyway. We’re lucky in the sense that we have one instance of Epic. I I know CMIOs who are dealing with multiple EMRs and multiple instances within their same organization. We feel really lucky that we only have one instance, and that’s been very intentional on our part. But we're used to having multiple vendors, and as you grow and make acquisitions, you're taking on different contracts anyway. So we're pretty used to that. I think managing multiple AI vendors is not any different than managing multiple lab vendors. 

HCI: This survey says only 11% currently work with a comprehensive AI partner who manages multiple AI use cases. But 72% said they they would prefer to operate under that kind of model. Would you prefer that kind of model?

Anderson: Well, I think that would be awesome. I don't know that we have that yet. It's not our EMR vendor, and it's not any of the vendors that we have worked with yet.

HCI: Have you seen the pace of of AI tools being introduced by Epic increase in the last year or two? 

Anderson: Absolutely. There's demand out there, and it’s definitely sped up. People are saying they’ve got to do this. I think the cream is going to come to the top eventually, and some things are going to settle out. But the pace at which things are coming out and things are being marketed and shown is accelerating.

We have conversations where one of our clinicians sees a shiny object out there, and it might work for this one tiny pillar of what we do. One of the things we focus on is taking a broader look. If you can make one step a lot better, you have to look at the step before and after and ask, is it making those other steps worse? Are we just shifting work one way or the other? That's what a lot of those solutions are. I can use AI or any technology to make this step right here a lot better. But what we're really doing is causing extra work on both ends, and that's not what we're looking for. You have to have a partner that understands that and can embed and implement into your workflows appropriately.

HCI: Are there expectations about measuring the return on investment for these AI innovations?

Anderson: That's one of the things that we talk a lot about in our transformation group. We can do lots of things, but what are we going to measure success by? If we can't prove that it's going to help a patient or a clinician, it's probably a non-starter. If we can't figure that part out, I don't even want to talk about it. 

I think you have to find both hard and soft ROIs. Are you going to drop costs in one place and maintain productivity? Are you going to increase productivity? Are you going to do better at revenue capture? Those are things that I think are very important, because these tools are not cheap; they are not free. We have to be able to pay for that. We also have to say, can we make things better? Can we be a better organization? Can we capture market share by being more patient-friendly and more clinician- friendly? There also is a component of what the landscape looks like. For instance, what if everybody has the ambient tools and we don’t? There is a component of what is expected versus what is novel and and I think the speed with which we're seeing that transition in AI from novelty to expectation is pretty significant.

HCI: Has HonorHealth hired a chief AI officer? Or thought about it?

Anderson: Not that I know of. They haven't asked me to do it. I think it comes down to how you view artificial intelligence. Is it something that needs to be managed or is it a tool that our team members use? You know, we don't have a chief Epic officer. People who work in it and that's their focus, they make it better. AI is another tool that we need to use. My personal opinion is that it is just a tool and and everyone has the opportunity to improve their own sphere with it.

HCI: So do you think as as health system operations mature in their use of AI, that title might go away?

Anderson: Absolutely. 100 percent. It's important, but I think it's going to go into the background soon.

HCI: Do you think that smaller community health systems are going to be at a disadvantage vs. the larger health systems when it comes to deploying and monitoring these AI tools? Are they going to be more reliant on the pace that their EHR vendor is going?

Anderson: Actually, I think it could be a good equalizer, because the one benefit that smaller systems have is that they'll have the nimbleness and they will have less of those institutional sunk costs. I think the bigger and more encumbered you are, probably the harder it is.