Journey Hub

What it does

In the Journey Hub App, we harness the unique capabilities of the Gemini API, specifically the Gemini Pro model, to conduct highly precise, metropolitan area-specific searches. When a user submits a request via the Flutter app, Firestore is immediately updated with their prompt. We then employ a vector index in Firestore, containing embeddings and associated intents, to perform a nearest neighbor search. This process identifies the user's intent, such as seeking cooling centers or medical help, by matching it with predefined intents within the app.

Gemini’s standout feature is its ability to deliver highly localized searches. Once the user's intent is identified, Gemini performs a focused, location-specific search within the user's metropolitan area, retrieving relevant data like cooling centers or medical facilities. This ensures that users receive timely, context-specific information tailored precisely to their geographic location.

The data retrieved through Gemini is then formatted using the Instructor Python library to ensure clarity and relevance before being presented to the user. This capability not only provides immediate, location-based assistance but also allows for ongoing refinement. As users provide feedback, we will continuously update the intent vectors to enhance the accuracy and relevance of future searches, making the app smarter and more responsive over time.