Hospital pharmacy has quietly become one of healthcare’s most data-intensive divisions. Pharmacy teams analyze controlled substance movement to detect drug diversion, optimize drug spending, monitor shortages to plan purchasing and clinical strategies, and ensure that every high-cost drug in decentralized locations is used before it expires. And the data behind these decisions lives across multiple systems – rarely in one place.
That complexity is exactly what’s driving pharmacy’s accelerating adoption of AI.
This Isn’t The Same AI Pharmacy Has Been Using
Pharmacy software has used machine learning, a subset of artificial intelligence, for years – surfacing anomalies in controlled substance transactions and predicting drug shortages before they are reported to the FDA or ASHP.
What’s different now is the ability to harness broader data sets across domains: generative AI instantly interprets data and answers questions in plain language, and agentic AI interprets data from multiple sources and takes action on that data without requiring a human to initiate every step.
Where AI Stands Today for Hospital Pharmacy Teams
Bluesight’s 2026 Hospital Pharmacy Operations Report revealed that nearly half (48.5%) of hospital pharmacy leaders report using AI in some capacity, and 23.7% use it daily – a sign it has moved past novelty into routine workflows. The most common applications cited are communications and presentations (31.8%), diversion detection (24.8%), and data synthesis across technology vendors (11.8%).
The Demand for Smarter Compliance Tools
Nearly a quarter of pharmacy leaders report using AI for diversion detection, but that definition can vary. For some, this could refer to machine learning that’s been built into diversion software for years. For others, it means something newer: using general-purpose tools like Microsoft Copilot or ChatGPT to help analyze and report on diversion data. In a separate Bluesight survey of diversion staff conducted in March 2026, 35% reported using third-party AI tools to analyze data – a signal of real demand.
It’s evident why this use case is popular. Diversion investigations routinely extend beyond 50 days, and recurring reports for diversion committees can consume nearly a full workday each time. But general-purpose AI tools weren’t built for this data, and routing sensitive information through external tools is a genuine concern for health systems. What diversion teams really need is a tool that reduces this investigative burden without creating new risk.
Bringing AI to the data – rather than the other way around – is the clear answer. Twenty health systems have adopted the Prism Assistant, a generative AI tool built directly into ControlCheck, Bluesight’s drug diversion monitoring software. Instead of cross-referencing reports or exporting to Excel, users ask questions in plain language and get structured, evidence-backed answers to questions like “How does Jane Smith’s waste rate compare to peers for the top anesthesia drugs?” or “Create an investigation packet on John Doe to bring to my diversion committee meeting.”
Early adopters, including Diversion Coordinators and Diversion Program Directors, report significant time savings across the tasks that previously consumed most of their day:
- Recurring reports finalized 96% faster
- Pre-investigation triage completed 90% faster
- Controlled substance variance analysis completed 97% faster
The Wave From Generative to Agentic
Generative AI still depends on someone asking the right question. The next shift for health systems is agentic AI – tools that initiate and complete work independently across multiple data sources. Instead of simply answering questions, this kind of system works more like a virtual teammate, delivering answers without waiting for a question.
For example, consider GPO prohibition for certain 340B covered entities. Today, pharmacy teams manually audit and collect information across their EMR, TPA, wholesaler portals, and self-managed spreadsheets to confirm their purchasing is compliant. Alternatively, an agentic system could execute the GPO prohibition workflow on a regular cadence without prompting:
- Pull 340B data on every qualifying transaction.
- Check every invoice for any drugs bought on a GPO contract.
- Cross-reference supply chain signals to surface relevant shortages or backorders.
- Write the Audit Report, a Note-to-File, or a self-disclosure letter to HRSA.
And this isn’t a futuristic concept. Bluesight Prism is being designed to connect data across procurement and inventory, shortage management, 340B compliance, diversion monitoring, and patient privacy monitoring – and then act on that data.
AI in pharmacy isn’t about replacing expertise, but rather about removing friction. The real opportunity is more than faster reports and shorter investigations. It’s the ability to move pharmacy teams out of manual coordination and into higher-value clinical and strategic work that strengthens patient safety, reinforces regulatory compliance, and protects financial performance.
As seen on Becker’s Hospital Review.