How Stanford Medicine uses Microsoft AI to transform oncology workflows

Microsoft‘s Azure AI Foundry is pioneering a sophisticated multi-agent AI collaboration system that fundamentally changes how AI applications are built and deployed. This breakthrough platform integrates coding, collaboration, and cloud hosting environments to enable complex orchestration of multiple AI agents working together—with Stanford Medicine’s tumor analysis system demonstrating its life-saving potential in healthcare settings. As major tech companies unveil competing advancements, Microsoft’s agent orchestration approach represents a significant evolution in AI deployment that will likely define the industry throughout 2025 and beyond.

The big picture: Microsoft’s Azure AI Foundry combines three essential components—code environment (Visual Studio), collaboration platform (GitHub), and cloud hosting (Azure)—to enable seamless multi-agent AI systems at massive scale.

• The platform supports over 70,000 customers processing 100 trillion tokens and generating billions of daily search queries, according to Microsoft.
• This approach represents the evolution of “ensemble learning” from traditional machine learning into sophisticated systems where multiple specialized AI agents collaborate on complex tasks.

Why this matters: Multi-agent AI systems distribute work among specialized AI components rather than relying on a single large neural network, potentially creating more efficient and effective solutions for complex problems.

• The platform’s agentic retrieval capabilities, always-on observability, and built-in trust features address key challenges in deploying collaborative AI systems.
• This represents a significant shift from simpler systems like generative adversarial networks toward more sophisticated orchestrated agent collaborations.

Real-world impact: Stanford Medicine is using Azure AI Foundry’s agent orchestration to revolutionize cancer care by automating personalized treatment planning.

• The system helps clinicians evaluate medical imaging, identify appropriate clinical trials, and create personalized patient timelines—tasks that previously required extensive manual clinical work.
• Only one percent of cancer patients currently receive these personalized plans, which are known to improve oncology outcomes.

What they’re saying: “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,” says Stanford School of Medicine CIO Dr. Mike Pfeffer.

• “The new healthcare agent orchestrator has the power to streamline this existing workflow by reducing fragmentation… and enables surfacing new insights from data elements that were challenging to search, such as trial eligibility criteria, treatment guidelines, and real-world evidence.”
• Stanford Health Care is researching how to implement this as “the first generative AI agent solution used in a production setting for real-world care for our cancer patients.”

Where we go from here: Azure AI Foundry’s agent system signals a broader industry shift toward agentic deployment as a primary focus for AI development in 2025 and 2026.

• The announcement comes amid major AI revelations from both Microsoft and Google this month, including Google’s new AI Mode in search and Gemini 2.5 model advancements.
• Developers can explore these capabilities through the Azure AI Foundry Agent Catalog, where they can fine-tune models, test them, and target agents to specific tasks.