Setu

Inspiration

Natural disasters leave people with more than physical damage. Survivors must manage insurance claims, government aid, nonprofit programs, missing documents, funding gaps, and strict deadlines, often while facing emotional and financial stress.

We wanted to build something that could do more than answer questions. We wanted an AI system that could actively organize the recovery process and guide families toward the next best step.

That idea became Setu.

What it does

Setu is an autonomous disaster recovery agent that helps survivors manage the full recovery process in one place.

Users can upload insurance policies, claim letters, receipts, identification documents, damage reports, and property photos. Setu organizes the information, identifies missing documents, tracks deadlines, estimates funding gaps, and creates a personalized recovery plan.

As the case changes, Setu updates the plan and recommends the next best action. It helps connect survivors with insurance resources, government assistance, and nonprofit programs while keeping their recovery progress clear and organized.

The name Setu means bridge in Hindi and Sanskrit. Our platform acts as a bridge between disaster survivors and the resources they need to recover.

How we built it

We built the frontend using React, Vite, TypeScript, and Tailwind CSS. The dashboard includes document management, funding gap analysis, recovery timelines, progress tracking, and live case updates.

We used FastAPI and Python for the backend and agent workflow. AWS supports the cloud infrastructure, Nexla helps connect and manage data, Zero.xyz supports autonomous actions, and Pomerium provides secure access across the platform.

The system continuously reviews the recovery case, creates a plan, takes the appropriate action, checks the result, and updates its next decision.

Challenges we ran into

One of the biggest challenges was creating an agent that could handle a recovery process that constantly changes. Claim decisions, available aid, missing documents, and deadlines can all change over time.

We also had to present a large amount of information without overwhelming the user. Since disaster survivors may already be under significant stress, we focused on making the interface calm, simple, and easy to understand.

Another challenge was connecting multiple technologies into one reliable workflow while keeping the frontend and backend clearly separated.

Accomplishments that we're proud of

We built a complete recovery workflow that brings documents, funding information, deadlines, aid programs, and recommended actions into one platform.

We created a personalized dashboard that changes as the recovery case develops.

We also designed the project around real human needs rather than building a general chatbot. Setu does not only provide information. It actively helps users understand what to do next.

What we learned

We learned that autonomous AI becomes more useful when it is connected to structured workflows, reliable information, and clear actions.

We also learned that building for people in stressful situations requires more than technical features. The design must create trust, reduce confusion, and clearly explain why an action is being recommended.

This project showed us that AI can support complex real world processes when it is designed around the needs of the people using it.

What's next for Setu

We plan to connect Setu with live insurance systems, FEMA and state assistance programs, and nonprofit databases.

Future versions could support multilingual assistance, proactive deadline notifications, mobile access, secure document submission, and direct communication with recovery organizations.

Our long term goal is to make Setu a trusted recovery companion that bridges the gap between disaster and stability, helping families recover with less confusion and greater confidence.

Built With

• amazon-web-services
• docker
• fastapi
• github
• llms
• nexla
• pomerium
• python
• react
• tailwind
• typescript
• vite
• zero.xyz