Hosted on behalf of the Association for Pathology Informatics (API), this interactive session demonstrates how emerging Generative AI techniques are transforming raw pathology data into actionable research and business intelligence systems. Pathology laboratories generate immense volumes of narrative text, yet extracting structured data elements and meaningful insights from unstructured narrative reports remains a key bottleneck in research and clinical operations endeavors. The first half of this session examines how Large Language Models (LLMs) can be leveraged to accurately, reliably, and efficiently parse unstructured pathology reports, transforming complex clinical descriptions into standardized, structured data elements without sacrificing accuracy or workflow speed. Building on this foundation, the second presentation explores the next evolution of AI integration: agentic AI systems. Attendees will learn how giving agentic AI tools safe, policy-compliant access to enterprise data warehouses enables dynamic querying, automates complex reporting, and unlocks deeper analytic efficiency across healthcare systems. Together, these presentation topics bridge the gap between unstructured clinical text and advanced enterprise analytics, providing attendees with practical strategies to implement generative and agentic AI tools safely within their own pathology informatics architecture.

Learning Objectives: 

1. Evaluate methodologies for utilizing Large Language Models (LLMs) to reliably and efficiently transform unstructured pathology report data into structured data formats.
2. Understand how agentic AI frameworks can be securely integrated with enterprise data warehouses to enhance data analytics efficiency and effectiveness.
3. Identify key governance, security, and integration practices required to safely deploy generative and agentic AI models within pathology enterprise ecosystems.

The growing use of artificial intelligence in pathology is creating new opportunities and responsibilities for pathologists, including the evaluation of AI-assisted tools in clinical practice. Responsible implementation is essential to ensuring reliable performance, equitable outcomes, effective workflow integration, and appropriate clinical accountability. Existing regulatory and accreditation frameworks provide a strong foundation for AI-based tests, yet laboratories continue to navigate how best to operationalize these tools across the test lifecycle. As implementation approaches evolve, greater consistency can help strengthen confidence in AI-assisted pathology practice. This session presents the CAP AI Implementation Guide, a practical, step-by-step framework that translates established laboratory principles into a structured approach for implementing AI-based tests and promoting trustworthy use of AI in patient care.

Learning Objectives:

• Describe the major stages of the AI-based test implementation in pathology.
• Identify the clinical, operational, and technical considerations needed to determine whether an AI-based test is suitable for pathology practice.
• Apply key reporting principles that support transparency, trust, explainability, generalizability, and clinical actionability in AI-assisted pathology practice.