What makes AI guidance trustworthy for pharmacists?
Some pharmacies view AI-enabled solutions as a quick way to streamline decision-making and manage heavy workloads. The panelists cautioned against adopting AI without first examining the medication intelligence behind it.
For Marshall, trustworthy AI guidance starts with where its content comes from. It should draw on a defined library of clinical content that experts maintain and review, and its recommendations should trace back to the primary source. Without a governed clinical AI solution available across the enterprise, individual pharmacists might turn to unsanctioned or open source shadow AI resources. These tools can create risks for the organization and patient care.
Governed, expert-curated AI can also give pharmacists more confidence that guidance is reviewed and updated as evidence changes, Marshall said. “What you might recommend 18 months ago might not be what you recommend 18 months from now.”
Having expert-reviewed guidance matters because pharmacists don't need more to sort through. “The last thing he or she needs is more information without guidance, without direction,” Marshall said. Lou DiPalma, Group Director of Central Fill at Walmart, holds every new tool to the same test. “If technology is creating more steps, more alerts, more work for the pharmacist, you have to question whether it's actually solving the right problem.”
Trusted AI clinical guidance can help ease that load by suggesting next steps and explaining the reasoning behind them. Pacete described tools that summarize a patient's information, flag what's changed, and prioritize who needs a call first. The pharmacist still decides whether a suggestion fits the patient. “We need to make sure the technology allows for the pharmacist to potentially choose outside of the AI tool, if the situation determines that that's necessary,” Pacete said.
Map and standardize before you automate with AI
The panelists warned against adding AI to unexamined processes. “We can spend a lot of money making a bad process happen even faster,” Pacete said. He recommends mapping and standardizing existing workflows first, beginning by asking why each step exists.
DiPalma takes a similar approach at Walmart with a framework he describes as eliminate, automate, simplify. His teams ask if a step is still needed, then whether it's a candidate for automation, and if not, how else to simplify it.
That thinking shapes Walmart's Central Fill network. It takes on repeatable dispensing steps such as pill counting, labeling, and verification, with pharmacists providing oversight. DiPalma said one facility in Maryland serves 16 states and Washington, D.C., each with its own board of pharmacy. Workflows built independently tend to break because they don't coordinate, he said. That's why his teams standardize the steps between each site.
