ProtectEd

Inspiration

After the recent tragic shooting in Nashville. we realized that the current system for reporting and responding to armed intruders in schools is a mess, albeit a crucial system, that is necessary to keep our students safe. We wondered if we could streamline this process with a combination of GPT models and programmable SMS and voice notifications.

What it does

ProtectEd is a mobile app that uses a combination of event-triggered SMS and voice calls combined with GPT-3.5 to administer communication between teachers, students, administrators, and authorities. At the beginning of the school year, teachers can add students to their classrooms, who will then be alerted via SMS if there ever is an emergency, as well as additional information about where their classroom is located so in the event of an emergency first responders can know where to go and who needs aid. To allow the teachers and staff to focus on protecting their students rather than communicating with first responders, a GPT-3.5 voice assistant will call dispatchers and answer questions about the event.

How we built it

The frontend UI is built in React Native allowing for compatibility with both iOS and Android. The backend databases and user authentication is managed by Google's Firebase. The calling and messaging systems are built with Twilio for their enterprise-grade features and OpenAI's GPT-3.5 model was used via their API for its superior conversational skills.

Challenges we ran into

Asynchronous event triggering can be a massive headache as dozens if not hundreds of texts and calls must be sent out and answered at once. Tweaking the GPT-3.5 prompt as well as managing the relationships between the UI, Firestore database, and Twilio call server was also difficult

Accomplishments that we're proud of

We are proud that we are finding new use cases for the groundbreaking technology that is GPT. As it can understand and interpret unstructured data, it can be extremely useful for reporting on real-time events if it's fed the right data and prompts.

What we learned

We learned that GPT-3 is capable of understanding and interpreting real-time data and answering questions about it in a way that people might not be able to do while panicking.

What's next for ProtectEd

We hope to reach out to our local community and see if solutions like ProtectEd can be implemented, not just for school shooters, but for all kinds of emergencies within our schools.

Built With

• firebase
• firestore
• google-cloud
• gpt
• openai
• react-native
• twilio