Historical Grant Analysis AI Infrastructure

Palm Health Foundation is a philanthropic organization that funds mini grants for individuals, small nonprofits, and small businesses. The organization needed a better way to review and learn from years of historical grant information spread across multiple systems.

In partnership with BeeNex, we developed an internal retrieval application that integrates with all three repositories and enables staff to ask natural-language questions about historical grant data through a conversational interface. The system was designed to support long-context analysis using Gemini on Vertex AI, while also preserving chat history so users could continue prior conversations and build on earlier findings. It also included authentication and programmatic user creation for foundation staff.

The foundation’s historical grant records were scattered across Box, Google Drive, and OneDrive/SharePoint, with information stored in many formats, including tracking spreadsheets, instructional documents, templates, interview summaries, and physical sign-in sheets. Because the data lived in separate repositories and inconsistent file types, staff had no efficient way to analyze it holistically or identify patterns that could inform future funding decisions.

Cross-Repository Retrieval Infrastructure For Historical Grant Analysis