Financial Triage AI

Inspiration

Across developing countries, millions struggle to prove financial hardship when applying for aid. Social workers spend hours manually reviewing complex, unstructured bank statements just to decide who qualifies for support. This process is slow, inconsistent, and often delays critical help to those in need. We wanted to build something that brings fairness and speed to this process—without compromising privacy. That’s how Financial Triage AI was born: to empower NGOs and aid organizations with faster, evidence-based, and ethical financial assessments that help fight poverty in real time.

What it does

Financial Triage AI is a privacy-first SaaS tool that analyzes bank statements using artificial intelligence to generate structured data, financial health scores, and poverty-risk insights—instantly. It allows social workers to upload a statement and receive a clean, auditable report showing income stability, spending trends, and vulnerability indicators. The system provides these insights without storing any personal data or using a database, ensuring complete data privacy while enabling NGOs to prioritize and assist those most in need quickly and fairly.

How we built it

We built Financial Triage AI using a combination of OCR, LangChain, and Large Language Models (LLMs) powered locally through Ollama to transform unstructured bank statements into structured, explainable financial data. The system, developed with Python, FastAPI, and Streamlit, uses Tesseract OCR and custom parsing rules to handle diverse statement formats, while a retrieval-augmented generation (RAG) engine enables intelligent Q&A with evidence citations from the extracted data. All processing is done in-memory without any database or data retention, ensuring complete privacy, transparency, and compliance with NGO ethical standards.

Challenges we ran into

Our main challenge was handling the wide variety of statement layouts and inconsistent formats across banks and regions. We also had to balance automation with interpretability—social workers need transparency, not just numbers. Designing a system that processes sensitive data without storing it required creative architectural choices, including ephemeral vector stores and instant data deletion. Making the interface both simple and secure was another key challenge, as the platform must be accessible even to non-technical users working in the field.

Accomplishments that we're proud of

We successfully created a working AI prototype that converts bank statements into structured, interpretable insights within seconds—without saving any user data. In test runs, our system reduced financial review time by up to 87%, enabling NGOs to process hundreds of additional aid cases per month. We’re proud that Financial Triage AI directly contributes to the United Nations Sustainable Development Goal 1: No Poverty, by helping organizations deliver faster, fairer, and data-driven assistance where it matters most.

What we learned

Through this project, we learned that ethical AI design is just as important as technical innovation. Building a privacy-first system pushed us to rethink how data should be handled responsibly in humanitarian contexts. We also discovered the importance of simplicity—social workers need clear, trustworthy insights, not complex dashboards. Our biggest takeaway is that technology can make aid delivery faster and fairer, but it must always protect the dignity and privacy of the people it serves.

What's next for Financial Triage AI

Next, we plan to partner with local NGOs to pilot Financial Triage AI in real-world aid assessment programs. We aim to expand our model to support multiple languages and integrate regional banking formats, enabling accessibility across different countries. Future updates will include offline mobile scanning for remote areas and customizable scoring systems tailored to each NGO’s criteria. Our long-term goal is to make Financial Triage AI a trusted global platform that accelerates poverty relief through secure, transparent, and ethical financial insights.