In early 2024 I volunteered with the American Red Cross to streamline an important but time‑consuming task: analysing receipts from hundreds of monthly donations and expenditures. Disaster response teams relied on manual review, which meant long hours and potential delays in reporting.
To tackle this challenge I built a generative AI agent using Azure PromptFlow. The system ingests PDF and image receipts, extracts structured data and uses a large language model to categorise and summarise spending. Key metrics are then loaded into a dashboard to provide an at‑a‑glance overview of funds deployed across different disaster response activities.
By automating receipt analysis we reduced manual effort by more than 85% and saved roughly 10 hours per week. The faster turnaround enabled teams to make quick decisions and ensure resources were allocated where they were needed most. This project reinforced my belief that modern AI tools, when paired with thoughtful design, can have a real impact on humanitarian efforts.