Agentic AI Enhances Cancer Care at Stanford University

Imagine a world where every cancer patient receives care that’s not just personalized, but orchestrated with the precision of a symphony—each specialist, each data point, each decision aligned seamlessly. That world is now emerging at Stanford University, where a new kind of AI teammate is helping clinicians tackle the ever-growing complexity of cancer care. This isn’t just another algorithm. This is agentic AI—a system where multiple AI agents work together, coordinating across disciplines, analyzing everything from pathology slides to clinical notes, and surfacing insights that were once buried in mountains of fragmented data. As someone who’s followed AI for years, I can tell you: this is the dawn of a fundamental shift in medicine[4].

Let’s face it: cancer care is complicated. Even at top institutions like Stanford, clinicians juggle patient histories, imaging results, genomic data, and treatment guidelines—often across different software platforms. The result? A workflow that’s as fragmented as a jigsaw puzzle missing half its pieces. Enter Microsoft’s agent orchestrator for cancer care management, now available in the Azure AI Foundry Agent Catalog as of May 24, 2025[1]. This platform allows developers to design, manage, and customize AI agents—think of them as digital assistants—that can work together to tackle tasks that would take hours, if not days, for humans to complete.

The Rise of Agentic AI in Medicine

Medicine is moving rapidly from using AI as a tool to deploying AI as a teammate—what’s being called “agentic AI.” In a recent Lancet article, Stanford’s James Zou described this as a “fundamental shift,” with AI agents not just providing recommendations, but actively collaborating with clinicians to make sense of complex data[4]. The goal? To democratize precision medicine, making it accessible within existing workflows for every patient.

At Stanford, this isn’t just theory. The healthcare agent orchestrator is already in use by cancer teams, alongside institutions like Johns Hopkins, Providence Genomics, Mass General Brigham, and the University of Wisconsin School of Medicine and Public Health[2]. According to Mike Pfeffer, M.D., chief information officer at Stanford Health Care and Stanford School of Medicine, “Stanford Medicine sees 4,000 tumor board patients a year, and our clinicians are already using foundation model-generated summaries in tumor board meetings today (via a PHI safe instance of GPT on Azure).”[2] The new orchestrator, he says, streamlines existing workflows by reducing fragmentation—no more endless copy-pasting between systems—and surfaces new insights from data elements that were previously hard to search, like trial eligibility criteria and real-world evidence[2].

How Agentic AI Works in Cancer Care

So, what does this look like in practice? Picture a tumor board meeting. Clinicians gather to discuss a patient’s case, but instead of sifting through disparate records, they have a unified interface powered by AI. The agent orchestrator coordinates multiple specialized AI agents: one analyzes imaging (DICOM files), another interprets pathology slides (whole-slide images), another processes genomic data, and yet another mines clinical notes from electronic health records[1]. These agents work together, generating summaries, highlighting potential treatment options, and flagging eligibility for clinical trials—all in real time.

Microsoft’s platform is modular, allowing for both general-purpose and specialized AI models. Importantly, it also supports open-source customization, so developers can tweak agents to fit their specific needs[1]. The orchestrator can integrate approved third-party agents as well. For example, Paige.ai’s “Alba” agent, which offers real-time, conversational pathology insights, is now available in preview—the first time a third-party agent has been integrated into the orchestrator[2].

Real-World Impact and Data

This isn’t just about convenience. The impact is measurable. For Stanford’s cancer teams, the agent orchestrator is expected to save hours of administrative work per patient, freeing up clinicians to focus on what matters most: patient care. By reducing fragmentation and surfacing actionable insights, the system has the potential to lower administrative roadblocks and transform care delivery[1][2]. Pfeffer notes that the orchestrator enables “surfacing new insights from data elements that were challenging to search, such as trial eligibility criteria, treatment guidelines, and real-world evidence.”[2]

Consider the numbers: Stanford Medicine’s tumor boards handle 4,000 patients a year[2]. If the orchestrator saves just one hour per patient, that’s 4,000 hours—over 166 days—of clinician time reclaimed annually. And that’s likely a conservative estimate. The real payoff comes from surfacing insights that might otherwise be missed, potentially leading to better outcomes and faster access to cutting-edge treatments.

Historical Context and Evolution

To appreciate how far we’ve come, it’s worth looking back. For years, AI in medicine was limited to narrow tasks: identifying tumors in imaging, predicting patient risk, or automating simple administrative chores. These were valuable, but they didn’t address the complexity of multidisciplinary cancer care. The breakthrough here is orchestration—getting multiple AI agents to collaborate, much like a team of specialists working together on a complex case.

Microsoft’s Azure AI Foundry platform is a key enabler, providing the infrastructure for designing, managing, and deploying these agents at scale[1]. This isn’t just about Microsoft, though. The broader trend is toward open ecosystems, where third-party developers and researchers can contribute their own agents, as seen with Paige.ai’s Alba[2]. This approach mirrors the collaborative spirit of modern medicine itself.

Future Implications and Potential Outcomes

The implications are profound. As clinical care complexity escalates, the need for intelligent orchestration will only grow. Agentic AI has the potential to become a standard part of care delivery, not just in cancer but across medicine. Imagine a future where every patient’s care is guided by a team of digital agents, each bringing specialized expertise to the table—analyzing data, surfacing insights, and even anticipating risks before they become crises.

This isn’t science fiction. It’s happening now, at Stanford and other leading institutions. The next step? Scaling up. If agentic AI can deliver on its promise, we could see a revolution in care delivery—one where precision medicine is truly accessible to all patients, not just those at elite centers.

Different Perspectives and Approaches

Not everyone is convinced, of course. Some worry about over-reliance on AI, or the risk of algorithmic bias. Others question whether these systems will truly save time, or just add another layer of complexity. But the early results are promising. At Stanford, clinicians are already using foundation model-generated summaries in tumor board meetings, and feedback has been positive[2]. The key, it seems, is integration—making sure the AI works within existing workflows, not as a bolt-on, but as a seamless part of the care team.

By the way, this isn’t just about cancer. The same principles could apply to any complex, multidisciplinary care setting—think neurology, cardiology, or rare diseases. The agent orchestrator architecture is designed to be flexible, adaptable to new specialties and new data types as they emerge[1].

Comparison Table: Traditional vs. Agentic AI in Cancer Care

Forward-Looking Insights

As someone who’s followed AI for years, I’m thinking that we’re on the cusp of something big. Agentic AI isn’t just a new tool—it’s a new paradigm. By orchestrating multiple AI agents, we can tackle the complexity of modern medicine head-on, delivering care that’s more precise, more efficient, and ultimately, more human. The future of cancer care—and, by extension, all of medicine—is collaborative, intelligent, and, yes, a little bit orchestrated.

Excerpt for Preview:

Stanford clinicians are using Microsoft’s agentic AI orchestrator to streamline cancer care, reduce administrative burdens, and unlock insights from complex, multidisciplinary data[2][1].

Conclusion

The story unfolding at Stanford is a glimpse into the future of healthcare. Agentic AI isn’t just about making tasks easier—it’s about transforming how care is delivered, enabling clinicians to focus on what they do best: caring for patients. With platforms like Microsoft’s Azure AI Foundry, and the integration of third-party agents like Paige.ai’s Alba, we’re seeing the emergence of a new era in medicine—one where AI is a true teammate, not just a tool. The road ahead is challenging, but the possibilities are exhilarating. As Pfeffer puts it, Stanford is excited to build “the first generative AI agent solution used in a production setting for real-world care for our cancer patients.”[2] And if the early results are any indication, this is just the beginning.

**