Rotabull Customer Response Suggester

Features

• Automated weekly data updates from Zendesk and README documentation
• AI-powered response generation using Anthropic's model
• Integration with Forge (https://forge-ml.com/) for efficient data storage and retrieval
• RESTful API endpoint for generating response suggestions
• Scheduled jobs for maintaining up-to-date support data
• PostgreSQL database for storing ticket and user information

Installation

1. Clone the repository:

   git clone [your-repository-url]
   cd rotabull-customer-response-suggester

2. Install dependencies:

   bun install

3. Set up environment variables in a .env file:

   ANTHROPIC_API_KEY=your_anthropic_api_key
   README_API_TOKEN=your_readme_api_token
   ZENDESK_API_TOKEN=your_zendesk_api_token
   ZENDESK_USER_EMAIL=your_zendesk_email
   ZENDESK_SUBDOMAIN=your_zendesk_subdomain
   FORGE_KEY=your_forge_key
   DATABASE_URL=your_database_url
   

Usage

To start the application:

bash bun run checkZendeskFirst.ts

API Endpoints

Generate Response Suggestion

POST /suggest

Request body:

{ "subject": "My subject", "requester": "[email protected]", "text_body": "I'm having an issue with XYZ...", "html_body": "

I'm having an issue with XYZ...

" }

Response: { "suggested_articles": [ "https://support.rotabull.com/docs/article-slug-1", "https://support.rotabull.com/docs/article-slug-2" ], "suggested_response": "AI-generated response text" }

Trigger Manual Job

GET /triggerManualJob

Manually triggers the data update job.

Development

The project uses TypeScript and is structured as follows:

• src/: Contains the main application logic
  • checkZendeskFirst.ts: Main Express application setup
  • fetchReadme.ts: Fetches and processes README documentation
  • fetchTickets.ts: Fetches and processes Zendesk tickets
  • fetchUsers.ts: Fetches and stores Zendesk user data
  • scheduledJobs.ts: Defines scheduled jobs for data updates
  • handler.ts: Serverless handler for the API
  • scheduledJobHandler.ts: Serverless handler for scheduled jobs
• prisma/: Database schema and migrations
• scripts/: Utility scripts for database operations
• forge/: Configuration for Forge ML integration

Key Components

1. Express Application: The main API is built using Express.js.
2. Prisma ORM: Used for database operations and schema management.
3. Forge ML: Integrated for efficient data storage and retrieval.
4. Serverless Framework: Used for deploying the application to AWS.
5. Cron Jobs: Implemented for regular data updates.

Deployment

The application is configured for deployment on AWS using Serverless Framework. To deploy:

1. Ensure you have the Serverless Framework installed and configured with your AWS credentials.
2. Run the deployment command:

   serverless deploy

Database Schema

The PostgreSQL database includes the following main tables:

• ForgeTicketCollection: Stores a reference to Forge collections for tickets
• ForgeDocsCollection: Stores a reference to Forge collections for documentation
• SupportDoc: Stores README documentation
• ZendeskTicket: Stores Zendesk ticket information
• ZendeskTicketComment: Stores comments associated with Zendesk tickets
• ZendeskUser: Stores Zendesk user information