Responder

Inspiration

With over 870,000 Indianapolis residents depending on effective emergency response, we saw a critical need for a solution that enables communities to rapidly report/ Or get help about hazards, coordinate responses, and triage emergencies using modern AI technologies. Inspired by the power of Multi AI Agents-style collaboration, Responder transforms public safety workflows into intelligent, coordinated systems.

What it does

Responder is an AI-powered chatbot system built with static expert agents that handle different aspects of public safety:

Hazard Reporter Agent – Gathers incident data from citizens in natural language.
Emergency Alert Dispatcher – Simulates sending alerts to appropriate responders or services.
Medical Triage Advisor – Provides basic emergency triage guidance using up-to-date medical knowledge.
Crisis Coordinator – Synthesizes all insights into a coordinated response plan.

Citizens interact with one interface; behind the scenes, agents collaborate and debate in real time to provide fast, multi-perspective answers.

How we built it

Frontend: Built with Next.js, TailwindCSS, and ShadCN for a responsive, accessible UI/UX.
Backend: FastAPI-based Python server that handles agent orchestration and queries to the Perplexity API.
Agent System: Static AI agents with professional prompts, each designed for a specific public safety task.
Markdown Rendering: Responses are rendered as rich markdown with clean formatting.
Cloud Deployment: App is deployable on any cloud platform and designed to scale.

Challenges we ran into

Handling agent coordination cleanly with markdown-formatted responses.
Ensuring user queries are understood accurately across diverse public safety domains.
Balancing the depth of AI response with speed and clarity for public use.

Accomplishments that we're proud of

Successfully integrated multiple AI agents working in parallel.
Delivered a full-stack, production-ready solution within hackathon constraints.
Created a system that could be deployed in real-world municipal infrastructure with minimal friction.

What we learned

Prompt engineering is critical for task-specific AI agents to be effective.
Coordination of AI agents simulates expert debate that can exceed single-bot answers.
UI/UX simplicity is key in emergency use cases.