Highmark Health and Google Cloud are revolutionizing the healthcare industry by integrating generative AI into their operations, streamlining medical claims and improving patient care. During a panel discussion at VentureBeat’s Transform 2025 conference, Google Cloud Platform Vice President and CTO Will Grannis and Richard Clarke, Highmark Health’s Senior Vice President and Chief Data and Analytics Officer, shared six key lessons from their collaborative project. Highmark Health, an integrated payer-provider system serving over 6 million members, is leveraging Google Cloud’s AI models and infrastructure to modernize legacy systems and enhance internal efficiency. The partnership focuses on platform engineering, treating AI as a foundational shift rather than just another technological layer. Clarke emphasized the importance of building flexible infrastructure early, which allowed Highmark to integrate even its COBOL-coded systems with cloud-based AI models. This approach has achieved up to 90% workload replication without disrupting existing processes, providing real-time insights into complex administrative tasks. Grannis stressed the need for a long-term vision, noting that preparing data and systems can take several years but is essential for AI to deliver meaningful outcomes. The two companies worked closely to ensure that more than 14,000 of Highmark’s over 40,000 employees are now actively using internal generative AI tools powered by Google Cloud’s Vertex AI and Gemini models. These tools are used for various tasks, from creating personalized member communications to automating claims processing. One notable example is the provider-side credentialing and contract verification process. Previously, this involved manual searches across multiple systems. With AI, these processes are automated, cross-checking requirements and generating tailored outputs complete with citations and context-specific recommendations. Clarke highlighted that Highmark does not simply deploy tools but actively trains employees and gathers user feedback, ensuring the technology is user-friendly and effective. The panelists also discussed the shift from chat-based interactions to multi-agent systems, which can complete complex tasks end-to-end. Grannis described these agents as coordinating multiple models, handling tasks from data retrieval to workflow execution. Highmark is piloting single-use agents for specific workflows, aiming to eventually embed them within backend systems for autonomous task performance. This reduces the need for multiple interfaces and enhances central control and reach. Both Grannis and Clarke advised adopting a task-first approach. They emphasized that businesses should define their outcomes before selecting the right AI models or architectures. Highmark uses Gemini 2.5 Pro for research-intensive queries and Gemini Flash for real-time interactions, along with classical deterministic models when appropriate. Google Cloud is investing in model-routing capabilities and open standards, such as the recent Agent Protocol initiative with the Linux Foundation, to promote interoperability and stability. For enterprise leaders considering AI integration, the panelists provided practical advice: Lay the foundation early: Invest in data readiness and system integration now, regardless of the timeline for full AI deployment. Avoid building foundational models: It is cost-prohibitive unless your business specifically focuses on model development. Instead, focus on orchestrating and fine-tuning existing models for specific tasks. Adopt a platform mindset: Centralize access to models and track usage to support experimentation while maintaining governance. Start with tasks, not tools: Define the desired outcome first, then choose the appropriate AI model or architecture. Measure and share: Internal adoption increases when employees see tangible results. Track usage, document success stories, and continuously update approved prompts and workflows. Design for action, not just information: Build AI systems that can trigger real-world actions safely and securely within existing systems. Clarke summarized the collaboration's success, stating that it is not about introducing flashy new features but about genuinely helping employees do their jobs better. The partnership between Highmark Health and Google Cloud serves as a valuable case study for enterprises across sectors looking to implement scalable, responsible, and highly usable AI solutions. industry insiders praised the innovative approach taken by Highmark Health and Google Cloud, noting that their emphasis on pragmatic integration and employee engagement sets a new standard for AI adoption in healthcare. The initiative demonstrates the potential of AI to transform administrative processes while maintaining clinical integrity and regulatory compliance. Both companies are committed to continuous improvement and sharing insights, making their collaboration a benchmark for future enterprise AI projects.