How Specialty Pharmacy Leaders Are Getting Started With AI
For a pharmacy team fielding calls all day or working through a prior authorization backlog, AI has an immediate appeal. However, evaluating it takes more work. You need to know how it will fit into your workflow, what your staff will need to review, and whether the results justify expanding beyond a pilot.
At NASP, Plenful’s SVP of Pharmacy Tim L’Hommedieu, PharmD moderated a panel with three pharmacy leaders working through those decisions: Haley O’Rourke, PharmD, CSP, Specialty Pharmacy Manager at Dartmouth Health; Rachel Chiou, PharmD, MS, BCPS, Clinical Manager of Specialty Home Delivery Pharmacy at UNC Health; and Anneliesa Hensley, Associate Director of Outpatient Pharmacy Services at The Ohio State University Wexner Medical Center.
The conversation focused on their experiences adopting AI, including what they’re testing and how they’re bringing their teams along.
The first use cases address familiar problems
"We found that we needed a more innovative solution to the volume of work that was coming through our team. So we began looking for a key solution, mostly to support the rapidly growing GLP-1 patient population. We knew that the volume was growing faster than our team could support.”
- Haley O’Rourke, PharmD, CSP, Specialty Pharmacy Manager, Dartmouth Health
Haley’s team at Dartmouth was facing volume that kept outpacing staff, particularly from the GLP-1 patient population. Their work includes benefit verification, prior authorization, and identifying potential referrals in chart notes. At UNC, the volume pressure was on the phones. AI-based call routing addresses call volume, an issue staff already wanted help managing. Conversation-driven routing offers a different experience from a traditional phone tree, with opportunities to take on additional call tasks as the technology develops.
Ohio State’s experience testing multiple prior authorization solutions adds a useful perspective on comparing tools before committing.
These are workflows where teams can identify the friction and assess whether something improves. They also require attention to the details of implementation. An authorization tool needs to work with the information and systems your team uses every day.
At Plenful, we’ve seen that in our work with Access Infusion Care, where understanding how coordinators reviewed and classified documents helped us tailor the technology to their workflow.
Decide what a successful AI pilot would show
The panel also recommended that before diving into the weeds with a pharmacy AI solution, start by defining the problem and the metrics you’ll use to evaluate it. That could mean time spent processing an authorization, the amount of rework, or the volume a team can handle.
Clear metrics give you a basis for comparing vendors and deciding whether to continue with an AI solution. Staff feedback helps explain the numbers: where the tool saves effort, where it adds steps, and what still needs attention.
A pilot or design partnership can give a pharmacy room to answer those questions while helping shape the product. Rachel Chiou raised that option on the panel: “There's definitely a design partner opportunity available.” It still requires time from your team. Agree on responsibilities, support, and evaluation criteria before getting started.
That same preparation applies beyond specialty pharmacy intake. In Tim’s 340B Coalition recap, he discusses mapping workflows and establishing baseline measures before introducing automation.
Give staff a role in testing and governance
Pharmacists and technicians need a way to influence how a tool works. Including them in testing helps surface exceptions and creates a feedback process grounded in daily practice.
It also makes conversations about changing responsibilities more specific. If AI takes on parts of an administrative workflow, explain which tasks will change, what staff will continue to review, and how the time saved could support patient-facing work.
On the panel, Rachel Chiou described how UNC worked with its AI call center vendor: “With our AI call center solution, it was very much through a partnership with a vendor investigating what our workflow was and getting through what their technology solutions were for our problems.”
For an AI governance committee, come prepared to explain the workflow just as clearly. What information does the system use? Where does a person review its output? How will the team identify and correct errors?
Tim and GK Brar, Plenful’s CTO, explored those questions in more detail during a recent webinar, including how experts control what a system learns.
Ask peers what AI implementation actually took
The panel ended with underscoring that vendor evaluation should include conversations with pharmacy teams already using the AI technology. They suggested asking about EHR integration, staff adoption, and support after launch.
Expect some work after go-live, too. Your team will need to review outputs and help the system learn your organization’s practices. Ask peers how they staffed that effort and how the experience changed over the first few months. Good AI tools will improve over time, and while they will take some work standing up, should quickly be producing measurable returns.
Make room to get started
Finding time to evaluate AI is difficult when the work it could help with already fills the day. The panel encouraged leaders to make that investment, starting with a defined workflow and a clear way to measure progress.
Check your organization’s existing vendor relationships and approval process. Identify a champion who can help navigate governance, and involve the pharmacists and technicians who know the workflow’s exceptions. A focused pilot can give your team the experience and evidence to decide where to go next.
