• NIH implemented H2O.ai's h2oGPTe to create a secure, air-gapped generative AI solution that reduced IT service tickets by up to 10,000 annually 
• The AI assistant serves 8,000 employees across 28 institutes without exposing data outside the NIH network 
• The solution integrates directly with SharePoint, providing source-linked answers from hundreds of policy documents

In an era where artificial intelligence is transforming industries across the board, government agencies with strict security requirements often find themselves unable to leverage these cutting-edge technologies. The United States National Institutes of Health (NIH), one of the world's foremost medical research agencies, faced this exact challenge, until they partnered with H2O.ai to implement a groundbreaking secure, air-gapped generative AI solution. 

The Challenge: Information Overload Across 28 Institutes

NIH manages hundreds of policies and business system guides supporting finance, procurement, travel, budget, and acquisitions across its 28 institutes. With such vast amounts of information scattered across numerous SharePoint documents, both staff and researchers struggled to quickly find the answers they needed.

The result? Support teams were inundated with repetitive "how-do-I" questions, taking valuable time away from more critical tasks. The existing search functionality was slow and inefficient, and maintaining current institutional knowledge consumed significant resources.

To address these challenges, NIH's Office of NIH Business Systems implemented a Business Assistant chatbot powered by H2O.ai's h2oGPTe technology.

Watch the video below to learn how this worked:

Why This Matters for the Public Sector

As the Director of ONBS at NIH noted: "By deploying H2O.ai's h2oGPTe inside our air-gapped environment, we delivered a secure Generative AI assistant that surfaces precise policy and procurement answers in seconds—freeing our team to concentrate on mission-critical improvements instead of repeating FAQs." 

Conclusion

The NIH's implementation of H2O.ai's secure generative AI solution demonstrates that even organizations with the strictest security requirements can harness the power of AI to transform operations. By creating an intelligent, context-aware assistant that operates entirely within their secure environment, NIH has not only improved efficiency and reduced support burdens but also established a model that other government agencies can follow. 

For more details on this implementation, check out the full case study.

This blog summary was written by h2oGPTe with a human in the loop.