Inspired by the devastating wildfires in currently ongoing Los Angeles, Pickle gets you out of a pickle. In times of disaster like these, people often struggle to communicate their location or status, and emergency responders face challenges in determining where help is most needed. Unlike traditional location-sharing apps, Pickle was made to have a low barrier to entry and usage, yet maintains privacy by only sharing your location during emergencies. We wanted to create a tool that bettered lives by offering emergency-activated location tracking, anonymous data for large-scale disaster response, and internet-connected AI-powered survival assistance, all without the boundaries of having to stay connected to the internet or cellular data. Our mission was to leverage a combination of modern and old technology (SMS MESSAGING?!) to save lives, improve rescue coordination, and lift people up on the worst day of their lives.
Pickle's unique workaround to internet connectivity allows it to function as the best emergency response app in evacuation scenarios. It forgoes the need for internet or owning a cellular subscription by leveraging the long-forgotten technology of SMS messaging. Most modern phone plans offer unlimited SMS messaging, and as such, Pickle can leverage a clever method to stay connected. By updating information in the app through exchanging encoded SMS messages, Pickle can allow you to keep up with your loved ones in these times of crisis. All tied together through Twilio's SMS services, you update the information, interact with Cohere agents (which are up-to-date on information through server-side web searches with Tavily!!), and verify with Auth0 Passwordless sign-in all in one place.
Pickle is an emergency app designed to keep people safe and informed during crises. It allows users to view their friends' real-time locations and safety statuses, such as safe, on-the-way, or in a pickle. Users can add friends, update their status, and access a map displaying all their friends' locations, (as well as contribute to a larger heatmap for first responders). The app also cleverly integrates with Twilio SMS, enabling users to update their location even without active Wi-Fi or cellular data, which not everyone may have access to in an emergency. Additionally, the app features an agent powered by Cohere AI and web searches, which users can access through SMS messages. The chatbot provides real-time, context-aware emergency assistance. The chatbot ensures users receive guidance on what to do in emergency scenarios, helping them navigate crises effectively regardless of connectivity and accessibility challenges.
Pickle was developed using a robust stack of technologies to ensure reliability and scalability. We used MongoDB as our database to store user data, locations, and status updates. The frontend was built with React and TypeScript to create an intuitive and interactive user interface. On the backend, we used Python with Flask to handle API requests and connect to external services. For authentication and secure login, we integrated Auth0's SMS passwordless login, balancing accessibility to the app and ease of use, as well as secure information. The chatbot feature was powered by Cohere, tied together by SMS functionality implemented through Twilio. Both the friend location map and first responder heatmap were built using Leaflet for interactive mapping. Finally, we hosted our servers on Heroku for seamless deployment and utilized GoDaddy to host our domain at picklehelp.us.
Due to the massive tech stack implemented by Pickle, we had to spend LOTS of time combing through documentation and working with new tools. Auth0's tokens to authorize the user through just phone numbers definitely took the most amount of time to figure out, shortly followed by implementing Cohere's V2 of chat. (As someone used to V1, this was very confusing.) The internet connectivity for the Cohere Agent that enabled it to be up-to-date was equally as challenging but completely worthwhile. Shockingly, the underlying SMS infrastructure Pickle relies on was one of the simplest steps, as Twilio and Heroku proved themselves to be easy to work with and understand.
The concept of Pickle is almost ingenious, and as a team, we're proud of being able to bring such a creative and unexplored idea to life. The idea of using SMS for information sharing in today's age has rarely been considered, yet we managed to persevere and create this project. Additionally, as stated above, the scale of Pickle's tech stack was a daunting challenge for the project. Initially, we didn't foresee the scale that this would grow to. As we continued building, we realized databases like MongoDB and hosting services like Heroku would have to be a part of the project in order to make it work. Accepting these challenges, we familiarized ourselves with these new tools for developers.
Although we're grateful for the experience of getting to learn so much in so little time. All three of our team members got almost zero sleep and began to work inefficiently as our eyelids struggled to stay awake against long walls of documentation. In the future, better arranging our workload to prioritize our physical/mental health (which would likely have a benefit on productivity as well!!)
In the future, we plan to expand Pickle with several new features. Enhancing the AI chatbot with multi-language support will make the app more accessible to users worldwide (likely, we'd prioritize natural-disaster-prone regions like Japan or the Phillippines). We also want to add voice and image support, improving accessibility for a wider range of users (we actually considered this, as Twilio offers voice call services, but ran out of time to fully flesh out the idea..) Additionally, we'd like to introduce automated alerts for more proactive safety measures. Evacuation route planning would be another key feature, helping users navigate optimal escape routes during specific disasters in coordination with authorities or their APIs.