SocialCalenderAI App

Summary

The app, designed to simplify scheduling, uses AI to match users' preferences, locations, and availability with nearby events and restaurants. By leveraging MongoDB for data storage and vector embeddings for better recommendations, the app provides personalized event and restaurant suggestions. Users can sign up, input their preferences, and have the app automatically schedule events with friends based on their availability. Integrated with LLMs and APIs like Google Places and Ticketmaster, it offers a seamless experience for discovering and organizing events. A cloud-based backend ensures smooth operation, while a visual calendar helps users track their upcoming engagements.

Takeaways

• 😀 The Social Calendar AI app simplifies scheduling by finding events or food places based on user preferences, availability, and location.
• 😀 The app uses two main databases: MongoDB for user, restaurant, and event data, and a vector database for embeddings of restaurant and event data.
• 😀 The app integrates APIs such as Rapid API (for restaurants) and Google Places API (for events), structuring and cleaning data before embedding it into the database.
• 😀 OpenAI's text embedding models (256 vectors) are used to match user input (e.g., cuisine preferences) with the most relevant restaurants and events.
• 😀 Users can provide preferences for food types, events, and availability, which are then used by the AI to suggest suitable activities and locations.
• 😀 The app features a simple agent framework that processes user input, filters relevant data based on location, and queries the vector database for recommendations.
• 😀 Vector Atlas search in MongoDB allows location-based filtering of events and restaurants, ensuring suggestions are nearby.
• 😀 The app's recommendation engine uses a retrieval-augmented pipeline (RAG) to rank and match events and restaurants to user preferences.
• 😀 The app includes a calendar feature to help users visualize their upcoming events and manage their schedules.
• 😀 The app provides notifications for friend requests and event invitations, allowing users to accept or reject them directly within the app.
• 😀 The backend is powered by MongoDB, Express.js, and AI models hosted on Amazon Bedrock, with Lambda functions handling server-side tasks.

Q & A

What is the main purpose of the Social Calendar AI app?

The app helps users schedule events such as meetings with friends or attending concerts by matching their availability and preferences with those of their friends. It scans local areas for events and food places to suggest suitable options.

How does the app use databases to store and retrieve data?

The app uses two main databases: a MongoDB database for storing user data, restaurant details, and event information, and a Vector DB (also within MongoDB) for embedding restaurant and event data. This allows for efficient matching and retrieval based on user preferences and interests.

What role do vector embeddings play in the app’s functionality?

Vector embeddings map restaurant and event data into a high-dimensional space, enabling the app to find similar events or restaurants based on user input. This is achieved using OpenAI's text embedding models, which allow the app to match users' preferences with the closest data in the database.

What types of external APIs does the app use for data gathering?

The app collects event data through the Google Events API and Ticketmaster API, while restaurant data is sourced from the RapidAPI service. These APIs provide structured data, which is then cleaned and processed for use within the app.

How does the app handle user input for scheduling events?

Users can input preferences for food, events, and their availability, which the system then uses to suggest events or restaurants. Users can also choose to let the app automatically decide the time, event, and people involved based on their input and that of their friends.

How does the system decide which events or restaurants to recommend?

The system uses a vector search within the MongoDB Vector DB to filter events and restaurants based on the user's location, preferences, and availability. The system also employs a Geo filter to ensure recommendations are relevant to the user’s local area.

What is the process of event scheduling after the system processes the user’s input?

Once the user inputs their preferences, the system queries the vector database for events and restaurants that match. After filtering for location and availability, the system either suggests an event or restaurant automatically or provides recommendations based on user input. It then schedules the event accordingly.

What features does the app offer for managing user preferences?

The app allows users to input and update their food cravings, event interests, and availability. They can also edit their profile, add or remove friends, and update their preferences. The app even provides a visual calendar to track upcoming events.

What is the significance of the calendar feature in the app?

The calendar feature provides users with a visual overview of their upcoming events. It helps users manage and view their schedules, offering a clear representation of scheduled events like concerts or dinners with friends.

Which technology stack is used to develop the app?

The app uses React Native for the frontend, Express.js for the backend, and MongoDB for database management. For the machine learning component, OpenAI’s text embedding models are used, and agents are managed using Crew AI. The app is hosted on Amazon Bedrock, with AWS Lambda for job assignments.