Inspiration
I went to the hospital with a friend, and we couldn’t arrange the money we needed. I knew of NGOs that could help, but applying meant gathering documents, figuring out what each organization required, and following up while we were already worried about treatment. That experience inspired Hope. We wanted to make it easier to ask for help and know what to do next.
What it does
Applicants email Hope with a request for medical financial assistance and attach documents such as a medical report, a hospital estimate, and an ID. Hope uses AI to review the documents and draft a reply that explains what’s missing or what needs clarification. Applicants can send additional or corrected documents in the same email thread. Reviewers can see the conversation and document findings together in a dashboard.
How we built it
We built a backend that links incoming emails and attachments to a case. AI reviews the documents, while the backend tracks what has been received and what still needs attention. We built a dashboard for reviewers and a Streamlit interface to test the workflow. We also integrated a PayPal payout step that uses the beneficiary’s email address once the case reaches the payment stage.
Challenges we ran into
Applications rarely arrive as a complete package. Someone might send a medical report first, an ID later, and then a corrected hospital estimate. We had to make sure Hope could recognize which document replaced an earlier one, preserve the case history, and avoid asking for something the applicant had already provided.
Accomplishments that we're proud of
We built a workflow that continues across multiple emails. An applicant can correct an estimate without restarting the application, and a reviewer can see what changed and why it matters. We’re proud that an ordinary email is enough to get started.
What we learned
AI can help flag inconsistencies, but reviewers need to see which document supports each finding. We also learned that the most helpful reply is often the most specific one. Telling an applicant exactly which estimate needs correcting gives them a clear action to take.
What's next for Hope
We want to pilot Hope with NGOs to learn how it fits their review process. We’d also like to support applications in more languages, improve how Hope handles unclear documents, and give applicants clearer updates as their case moves from submission to decision.
Built With
- ag-grid
- alembic
- fastapi
- google-gemini
- imap
- next.js
- openrouter
- paypal
- postgresql
- python
- react
- smtp
- sqlalchemy
- streamlit
- tailwind-css
- typescript
- vite
