SoS.AI.Indiana

Inspiration

Our inspiration for SOS AI stemmed from recognizing critical communication gaps in public safety, particularly within local communities. Traditional reporting methods can often be slow, cumbersome, or inaccessible in urgent situations. We saw a clear need for a more immediate, user-friendly, and empowering tool that could streamline hazard reporting, disseminate vital emergency alerts swiftly, and provide reliable public safety information, directly benefiting Indiana communities by fostering proactive citizen engagement in maintaining collective safety.

What it does

SOS AI is a simulated Public Safety Chatbot designed to be a direct, conversational interface between citizens and public safety services. It empowers users to:

Report Hazards: Guides users through a step-by-step process to report various hazards (e.g., broken streetlights, potholes, fallen trees), collecting necessary details like type, location, and additional information.

Access Emergency Alerts: Provides instant, up-to-date simulated emergency alerts and advisories relevant to the community, covering weather, traffic, or other critical incidents.

General Public Safety Assistance: Responds to general inquiries about public safety, offering helpful information and guidance. The app aims to make public safety engagement as simple as having a conversation.

How we built it

SOS AI was built as an interactive web application using React for the frontend, providing a dynamic and responsive user interface. Tailwind CSS was utilized for rapid and consistent styling, ensuring a clean and intuitive user experience. The core conversational intelligence of the chatbot, including intent recognition and generating natural language responses for both general queries and the structured hazard reporting flow, is powered by the Gemini API (gemini-2.0-flash). Hazard reports and emergency alerts are handled using mock data and rule-based logic within the application to simulate real-world scenarios.

Challenges we ran into

One primary challenge was simulating accurate intent interpretation within a public safety context, where precision is paramount. Ensuring the chatbot correctly understood nuanced user requests, especially during the multi-step hazard reporting process, required careful design of conversational flows and prompt engineering for the underlying AI. Another challenge was managing conversational state effectively to guide users through structured tasks (like reporting a hazard) while remaining flexible enough for general inquiries. Finally, creating a convincing "real-time" experience for alerts and reports without actual backend integrations posed an interesting design hurdle.

Accomplishments that we're proud of

We are particularly proud of successfully implementing a seamless and intuitive conversational flow that guides users through essential public safety tasks. The ability to leverage the Gemini API to generate dynamic, context-aware responses, including explanations and confirmations, is a significant achievement. We are also proud of the clean, accessible user interface that makes public safety engagement approachable for a broad audience, demonstrating how advanced AI can be applied to create practical, impactful solutions for communities.

What we learned

Through building SOS AI, we learned invaluable lessons about the nuances of natural language processing in critical applications, where clarity and accuracy directly impact user safety. We gained deeper insights into designing user experiences for diverse demographics, emphasizing simplicity and directness over complexity. We also reinforced the importance of structured conversational design for tasks requiring specific information, even when powered by flexible LLMs. The project highlighted the immense potential for AI to bridge communication gaps in public safety, provided it's designed with robust, human-centered principles.

What's next for SOS AI

Our future plans for SOS AI include:

Integration with Real Public Safety APIs: Connecting the simulation to actual local government and emergency service APIs for live alerts and direct hazard submission.

Advanced Location Services: Implementing precise geolocation for hazard reporting and personalized, location-specific emergency alerts.

Multilingual Support: Expanding the chatbot's capabilities to serve diverse linguistic communities within Indiana.

Media Uploads: Allowing users to upload images or videos alongside hazard reports.

Predictive Analytics (Simulated): Exploring how AI could predict potential hazards based on historical data patterns.

Community Feedback Loop: Designing a system for real-time feedback on reported hazards and alert effectiveness.

Built With

• all