Agentic AI applications for real-world issues

We aim to build Agentic AI applications that solve India’s real-world problems by automating citizen support, education, and resource management using IBM’s open-source ADK.

Comment

Inspiration We were inspired by the idea that AI should not only think but also act. Many problems in society—such as women’s safety, healthcare, disaster response, and education—still lack tools that can respond intelligently in real time. Our inspiration came from real-life incidents where quick action could have saved lives. That is why we set out to build an Agentic AI Safety Assistant that can autonomously take steps in emergencies, starting with a Women Safety Agent.

What It Does • Detects emergency situations when the user presses a panic/help button. • Shares live GPS location with emergency contacts. • Finds and displays the nearest police stations using Google Maps API. • Sends an SOS alert via SMS through Fast2SMS API. • Works with a simple, clean UI to make it usable in panic situations.

How We Built It • Frontend: HTML, CSS, JavaScript for the user interface. • Backend: Python Flask server to handle API requests. • APIs Integrated: o Google Maps API for location + nearby services. o Fast2SMS API for instant emergency alerts. • Agentic AI Logic: An if-then autonomous pipeline where the AI takes action after the help signal is triggered. In simple terms: User presses Help  ⇒  Trigger AI Agent  ⇒  {Share Location+Send SMS+Guide User}\text{User presses Help} \; \Rightarrow \; \text{Trigger AI Agent} \; \Rightarrow \; {\text{Share Location} + \text{Send SMS} + \text{Guide User}}User presses Help⇒Trigger AI Agent⇒{Share Location+Send SMS+Guide User}

Challenges We Ran Into • Configuring and securing API keys properly. • Handling CORS errors between frontend and backend. • Ensuring the system runs fast and reliably in emergencies. • Designing the app for panic usability (one-click help, minimal steps).

Accomplishments That We’re Proud Of • Built a working prototype of an emergency safety system. • Successfully integrated maps + SMS + Flask backend within the hackathon timeline. • Designed our first Agentic AI system that doesn’t just give information but actually acts. • Learned to think about real-world usability of AI beyond experiments.

What We Learned • Agentic AI is about autonomy and real action, not just intelligence. • Importance of fast, reliable decision-making in safety-critical apps. • How to combine location-based services with real-time communications. • How collaboration and team effort speed up problem-solving.

What’s Next • Add voice activation (trigger SOS by saying a secret word like “Help me”). • Build a cross-platform mobile app for easier accessibility. • Extend the same Agentic AI framework to healthcare, disaster relief, and education. • Integrate predictive safety analytics (warning users before they enter unsafe areas).

Share this project:

Updates