Discreat

Inspiration

As women, we’ve grown up hearing chilling stories of domestic abuse within our communities. A common question often asked is, “Why not just call the authorities?” But research shows it’s not that simple. One of the biggest challenges victims face is the lack of discreet reporting methods — a problem our solution tackles head-on.

What it does

Discreet Help Button

Instead of building a standalone app — which risks losing its discretion once publicized — we partner with food delivery platforms like UberEats, DoorDash, and Foodpanda to embed a hidden help button directly within their checkout pages. It takes less than a second to trigger, and it leaves no visible trace in the app. Once activated, it immediately alerts the nearest police department based on the user’s delivery address.

Real-time Monitoring

While authorities are en route, our smart dashboard tracks their response time and routes in real time — allowing us to monitor how efficiently each case is attended to. Additionally, we use sentiment analysis to automatically update the case status to “resolved” when a police officer confirms resolution through the hotline.

Data Collection

In our database, we collect detailed case data which includes:

City and country
Time and date of activation
Responding police department
Response duration

Over time, we aggregate this data to measure average response times, detect high-incident areas, and analyze which food delivery platforms are generating the most reports.

Downloadable Datasets

We believe data is power. To reduce prosecution drop-offs due to a lack of evidence, users can download a timestamped report of their case history — empowering them with clear, verifiable documentation to support their legal process.

How we built it

Our key data models were reports and event logs, and with MongoDB Atlas and Google Maps API, we could encode location information in each. Then, with MongoDB Atlas geospatial queries, we could find all reports within a certain area, and display trails of location changes with Google Maps. For our chat feature, we used Cerebras API with a Llama model to gauge police responses and automatically mark reports as resolved. The frontend was built with React and Tailwind CSS.

Challenges we ran into

At first, we were stuck on how best to encode and track location, whether it was using address/coordinates, or even how to connect it all. We realized that Google Maps API had a built in Places API that had latitude and longitude for any addresses, and were able to connect it to the Point datatype in MongoDB Atlas. That took a lot of planning, though! We also spent the most time integrating Cerebras API and sentiment analysis, as we had to think deeply about what feature of our product it would bring the most benefit to. We ended up deciding that analyzing responder messages would provide a key feature of automatically marking records as resolved.

Accomplishments that we're proud of

We gained hands-on experience with a range of new tech tools, which we now feel much more confident using. More importantly, we learned the value of clear role allocation and trusting our teammates to execute on a shared vision. By leaning into each other's strengths and collaborating effectively, we were able to complete tasks efficiently and stay aligned throughout the project.

What's next for Discreat

We plan to reach out to food delivery service providers to explore partnership opportunities and bring our embedded help button to life. Additionally, we will collaborate with domestic abuse centers and police departments to ensure our solution is aligned with their needs and workflows.