TLRT: From Manual to Agentic: Building Trustworthy AI Workflows in Healthcare

Summary

The collision of accelerating technology and crippling workforce shortages has catapulted agentic AI from a theoretical concept into an urgent strategic necessity for healthcare leaders. Agentic AI, capable of reasoning, planning, and taking autonomous action, promises to fundamentally rewire how care is delivered and supported. Yet, realizing this potential requires navigating significant hurdles related to trust, data quality, and organizational willingness to embrace radical change.

To explore this pivotal moment, CHIME convened a Thought Leadership Roundtable, sponsored by Innovaccer, as part of the CHIME Innovation Summit, Moon Shots in Digital Health Innovation, hosted by CHRISTUS Health in Texas. This session gathered CIOs and clinical informatics leaders. Their unfiltered discussion focused on pioneering practical entry points for advanced AI, building robust guardrails, and managing the profound cultural shift required to operationalize truly intelligent workflows.

Russ Branzell, CHIME President and CEO, ignited the discussion by citing the industry’s central dilemma: “The government really wants to get their hands around AI, but the cycle of change is now measured in weeks, not months or years. By the time they understand today’s concept, it’s already old.”

The session was led by Russ Branzell, CHIME’s President and CEO, and sponsored by Innovaccer, whose President of Platform Products, Ashish Singh, joined the discussion.

The digital health leaders participating in the roundtable were:

• Chani Cordero, Chief Digital Officer, Brooke Army Medical Center
• Stuart James, Vice President, Chief Operations Officer & Deputy CIO, CHRISTUS Health
• Thomas Abraham, System Director, Innovation Integration and Development, CHRISTUS Health
• Tim Pugsley, Chief Information Officer, Prairie Lakes Healthcare System
• Katherine Taylor-Pearson, President Elect, ANIA

‍

Why Agentic AI Is Gaining Strategic Momentum

For many organizations, the pressure to adopt agentic workflows is driven by an economic reality: they must scale services without proportional increases in staff or physical infrastructure. The potential for AI to enhance quality while relieving persistent human burden makes its deployment a matter of organizational survivability.

The most promising opportunities identified centered on high-friction administrative tasks and complex clinical review processes.

‍

Eliminating Delays and Documentation Burden

“At its core, the motivation for many participants boiled down to survival in the face of labor shortages. Tim Pugsley, CIO, Prairie Lakes Healthcare System, forcefully argued this reality: “Automation equals survivability,” Pugsley said, stressing the importance of getting work done despite reduced labor. He highlighted the pre-authorization process as a prime candidate for immediate agentic intervention, noting that in rural areas covering thousands of square miles, taking multiple days to complete a prior authorization requires patients to come back another day for care. “This leads to long drive times, more time off work, and other personal disruptions that could be avoided with a streamlined process. This is why automation has been important to our operations and care.”

Similarly, the administrative burden clogging up workflows needs to be addressed. Thomas

Abraham, System Director, Innovation Integration and Development, CHRISTUS Health, Underscored the immense potential to automate the clinical clerical staff who “spend a lot of time reading things now, ” noting that automation allows this information to be integrated directly into the EHR so staff “can just get that information right in their EHR, and take action from their EHR.”

The nursing profession stands to benefit significantly, as Katherine Taylor-Pearson, President Elect, ANIA, championed, “AI can be a powerful game changer in reducing documentation burden and mitigating staffing shortages.” She added that one of the greatest benefits lies in optimizing data flow and interoperability. “The time nurses spend searching for health records or critical information—such as during medication reconciliation—could be dramatically reduced. Imagine if, upon a patient’s arrival, you could instantly access their complete health history, their entire story, to support informed and timely care decisions.”

‍

Shifting to Wellness and Proactive Care

The true value of AI lies in driving a systemic shift from an illness-based model to one focused on prevention and wellness. Stuart James, Vice President, Chief Operations Officer & Deputy CIO, CHRISTUS Health, illuminated this vision through the concept of “collective intelligence,” stating, “The holy grail is when the AI and the human are working together collectively to make decisions that would not have been possible by working alone.”

Branzell offered a sweeping glimpse of the future state: “We’ll get rid of keyboards, mice,

and screens — AI will follow us into the room like a hologram assistant.” This collaboration is necessary for achieving bio-individuality in care. Chani Cordero, Chief Digital Officer, Brooke Army Medical Center, focused on the highest-impact clinical applications, stating that the ultimate goal is for AI to “be able to solve and detect certain types of diseases, cancers, and or conditions.” She also proposed that a major key to success is figuring out “how to get the patient to care about their care” by designing systems that motivate healthy choices.

‍

Getting From Vision To Reality: Preparing For Implementation

The participants acknowledged that the industry’s innate risk aversion and lack of data maturity pose substantial barriers to adoption. The transition to agentic workflows requires a deeper commitment to governance, culture, and foundational data work.

‍

The Risk of Automating Chaos

A primary concern is that rushing AI implementation without first cleaning data and standardizing processes will only magnify existing inefficiencies. James issued a critical warning: “If you automate something bad, you run the risk of being very good at being bad.” This immaturity in thinking was echoed by Abraham: “Everyone wants AI, but most don’t even know what they want — it’s adoption without rationalization.” This sentiment was echoed by Ashish Singh, President of Platform Products at Innovaccer, who noted the problem of “garbage and garbage out” when the foundational data is not solid.

‍

Overcoming the Trust Barrier

The emotional and psychological hurdles surrounding AI remain significant. Staff often fixate on the failures of AI while ignoring human error. James offered a powerful statistical comparison: “We have a much higher bar for AI. We trust humans who were 76% accurate but reject AI that’s 86% accurate.” Pugsley observed this cognitive bias, pointing out the human preference for familiar error: “If you want to increase the risk for error in a process, put a human in line.”

Acknowledging and accepting the imperfections on both the artificial and human sides, healthcare can better focus on the critical point: while AI drives efficiency, the ultimate goal remains preserving and re-energizing the human role — ensuring that this new era of technology leads back to the ‘holy grail’ of collective intelligence, where human expertise and judgment are essential for optimal outcomes.

For clinical leaders, data errors extend the risk to patient safety and liability. Taylor-Pearson voiced a profound concern about AI-generated data and patient safety, explaining, “What keeps me up at night is the possibility of an AI-driven error. For example, if an incorrect medication dose were generated, and a clinician administered it, the outcome could be devastating. That’s the human cost of data integrity, and why accuracy and oversight must remain our highest priorities.”

She advocated for developing the steps necessary to build clinical confidence: “We need more workflow analysis, more cognitive use cases, and more nursing-specific research.” Furthermore, to get staff onboard, she insisted: “We have to walk their shift, show them how AI helps them manage a crisis, not just tell them.”

The most effective way to normalize AI use is by introducing the tools earlier in staff careers, including during schooling and training. Taylor-Pearson emphasized the importance of preparing future clinicians to collaborate effectively with AI technologies. “I encourage my students to actively engage with AI tools so they can begin developing critical analytical skills,” she explained, underscoring the need to evaluate accuracy and identify potential hallucinations or errors. “I want them to question the output, not to distrust technology but to use it thoughtfully and responsibly.”

Security also stands out as an existential risk. Cordero stressed that over-reliance on technology without a Plan B can be disastrous in a major cyber event. She cautioned against the loss of fundamental resilience, noting: “In a major event that knocks out digital prescribing, the ability to get medications for patients can be severely delayed or even halted. Many staff have never seen a handwritten script in years or at all.” She encapsulated the danger clearly: “The more we divide the system with new connections and tools, the more we use them, and the more vulnerable we become.”

‍

Defining Success and Driving Change

The participants acknowledged that the current U.S. healthcare economic model reliant on a demographic balance that has fundamentally shifted is unsustainable. The stakes are immense.

Branzell framed the crisis as existential: Healthcare’s funding model is based on outdated demographics and is no longer sustainable, a scenario threatening a profound fiscal collapse. He concluded the only way out is radical change: “For the first time, the equation doesn’t work. There must be a different solution, and maybe AI is the answer.”

Where will such a disruption originate? Singh reasoned disruptions come from two vectors, necessity and innovation. “On one hand, there is a dire need or challenge to meet, such as the capacity and demand challenge in healthcare,” he explained. “On the other hand, there is rapid advancement of technology, such as AI, ahead of disruption that increasingly promises to revolutionize processes and outcomes, including healthcare.”

The leaders agreed that a strategy built on small, proven victories must precede large-scale disruption. Success is not just measured in cost savings, but also in the simplification of the entire healthcare journey.

Singh distilled the overarching success metric to a single word: simplicity. “We need to make healthcare simple from a patient perspective and from a provider perspective,” he said. “A patient should be able to get an appointment booked in one simple call and not have to wait six months to be seen. And providers should have all the information at their fingertips to make accurate and timely assessments and recommendations.”

Abraham articulated the necessity of phased rollout: “I think really small wins get us levels of success. It proves that it works.” Taylor-Pearson agreed on the importance of achieving measurable results but offered a cautionary note: “Small wins validate progress, but we cannot afford to move too slowly. If we do, we risk being left behind.”

The fundamental challenge remains rewiring the organization’s mindset. James tackled the structural economic issues that relentlessly stifle progress: “With limited exceptions, the healthcare system is designed around episodic care for people who are sick rather than continuously managing wellness. Until this changes, our industry will struggle to innovate.” He delivered a potent caution that incumbents who don’t think outside the box “are at risk of being massively disrupted.” He added, “Healthcare has been surprisingly resistant to disruption, but we’re not immune. We could get Blockbuster-ed.”

Finding the right route and speed through the minefield will require a profound cultural shift. Pugsley concluded: “We have always supported the status quo; that is how we have served our communities for 100 years. We must rewire our organizations to think differently for a chance to thrive.”

James summarized the required “think big, be brave” mind shift: “Everyone who ever changed the world was crazy until they weren’t.”

‍

Vision Without Limits: Agentic AI Wish List

To close the discussion, participants dreamed big about which single accomplishment involving agentic AI workflow would have the greatest impact. The leaders’ wishes spanned structural reform, cultural acceptance, and patient empowerment:

• James advocated for a dynamic system of continuous innovation: “I would create a vehicle for continuous innovation. This would involve people whose job it is to come up with five or six ways a month that we can try to improve healthcare.”
• Branzell dared to imagine a world where knowledge is instantly shared: “Anybody collecting government money of any kind should not be able to hold technology or best practice and care as a competitive advantage.”
• Singh wanted an AI that would focus on human health as a whole: “Keep people healthy. Look at humans more holistically, not just as patients.”
• Cordero crystallized the ultimate goal as improved patient accountability: “We need to figure out how to make people care about their care. We must find a way to build accountability — like the Aldi cart, where you only get your quarter back if you return it.”
• Pugsley sought to empower the consumer: “Access to care needs to be convenient for the consumer. We must also empower our communities to have the information they need to make good decisions about their health.”
• Abraham focused on ease of use and clarity: Make things easier. “Use the technology and platforms we have to serve people that like to make it easier.”
• Taylor-Pearson championed the principle of universal care access, envisioning a future where “everyone has access to quality care—unconstrained by social, economic, or geographic barriers.” She emphasized that this vision must include rural America, where limited connectivity, provider shortages, and infrastructure gaps continue to challenge equitable access to healthcare and technology-driven innovation.

‍

Conclusion: The Mandate For Proactive Disruption

The discussion concluded that while agentic AI is poised to deliver meaningful change, the technology is merely the vehicle. The real revolution is the organizational willingness to shed ingrained risk aversion, demand best practices, and use intelligence to solve the economic and societal crisis of a fragmented care system. Success will be determined not by the size of the initial pilot, but by the courage of leaders to proactively disrupt their own systems before external pressures and unsustainable economic realities disrupt them first.