VexaSearch

VexaSearch is a simple AI-powered search application designed to determine the actions to perform based on a function call.

Currently, the application is able to perform the following tasks:

Generate images
Search the internet
Retrieve information about a specific URL

Models currently used:

firefunction-v1 (for function calls)
Mixtral for text generation
stable-diffusion-xl-1024-v1-0 (for image generation)
nomic-ai/nomic-embed-text-v1.5 (for RAG)

Prerequisites

Supabase account
Firework account
Google Custom Search Engine API key and Search Engine ID
Bing Search API key

Supabase

Create a new project on Supabase
Create a Search table with the following columns:

id (Primary Key)
response (Text)
links (Text Array)
query (Text)
slug (Text)
created_at (Timestamp)

CREATE TABLE Search (
  id SERIAL PRIMARY KEY,
  response TEXT,
  links TEXT[],
  query TEXT,
  slug TEXT,
  created_at TIMESTAMP DEFAULT now()
);

Create a Query table with the following columns:

id (Primary Key)
query (Text)
slug (Text)

CREATE TABLE Query (
  id SERIAL PRIMARY KEY,
  query TEXT,
  slug TEXT
);

Installation

Clone the repository

git clone https://github.com/n4ze3m/vexasearch.git

Copy the .env.example file to .env and fill in the required information

cp .env.example .env

Open the .env file and fill in the required information

Install the required packages

npm install

Start the application

npm run dev

This will start the application on http://localhost:3000

or you can start the application using Docker

docker-compose up

This will start the application on http://localhost:3000

Support

If you like the project and want to support it, you can buy me a coffee. It will help me to keep working on the project.