AI Rivet Search API

This repository contains an Express.js server that serves as an endpoint to interact with a Rivet graph to process queries and return a list of products using MongoDB. This is a complimintery repository for Article : Building AI Graphs with Rivet and MongoDB Atlas Vector Search to Power AI Applications. Additionally, there are 2 simple graphs that can be run through Rivet's UI tool.

Overview

The server uses Rivet's capabilities to run a graph that takes user queries, processes them, and executes a search against a MongoDB database to retrieve relevant product information.

Getting Started

Prerequisites

• Node.js
• npm or yarn
• MongoDB Atlas
• Access to Rivet and the relevant plugins

Complete steps 1-5 from the AI-Shop github tutorial : https://github.com/mongodb-developer/ai-shop

Installation

To run the Rivet Graphs from the UI follow the instructions to install Rivet and Setup OpenAI key and the MongoDB Plugin. Once installed import the project file into Rivet.

To run the server side please follow:

1. Clone the repository to your local machine.
2. Run npm install to install all the dependencies.
3. Make sure to set up your .env file with the following variables:
   • DB_URL: Your MongoDB connection string.
   • GRAPH_ID: The ID of your Rivet graph ("puzj17gj_xu56GUlr7164").
   • OPEN_AI_KEY: Your OpenAI API key.
   • RIVET_MONGODB_CONNECTION_STRING: The MongoDB connection string for the Rivet plugin.

Running the Server

Execute npm run server to start the server. It will listen on the defined PORT (default is 3000).

Usage

Send a POST request to /aiRivetSearch with a JSON body containing a query. The server will process this query and return a list of products.

Example request:

POST /aiRivetSearch HTTP/1.1
Host: localhost:3000
Content-Type: application/json

{
   // eg. "Banana and Milk" or "Lasagne reciepe"
    "query": "Your search query"
}

curl -X POST http://localhost:3000/aiRivetSearch \
-H "Content-Type: application/json" \
-d '{"query": "Banana and Milk"}'

You will receive a response with the search results, the original search list, and the pipeline used for the MongoDB aggregation.

Features

• Integration with Rivet for graph processing.
• MongoDB aggregation for efficient data retrieval.
• CORS enabled for cross-origin requests.
• Body parsing for easy access to request payload.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

If you have any questions or comments, please open an issue on this repository.