AI Designer API Server Specifications & Deployment Guide

Overview

AI Designer API is deployed as a standalone FastAPI application with Gunicorn/Uvicorn that communicates with Joget DX instances through the AI Designer plugin. This guide provides server specifications and deployment requirements.

Server Specifications

Minimum Requirements

Recommended Requirements (Production)

Deployment Options

Architecture

AI Designer API can be deployed in two configurations:

• Separate Server (Recommended for Production):

• • Dedicated server for AI Designer API
  • Joget DX instance on a separate server
  • AI Designer plugin communicates via HTTP/HTTPS

• Same Server (Development/Testing):

• • Both the AI Designer API and Joget DX are on the same server
  • Requires additional resources (combine specs)
  • Suitable for smaller deployments or testing

Communication Flow

Joget DX Instance → AI Designer Plugin → HTTP/HTTPS → AI Designer API Server

Deployment Methods

Container Deployment (Recommended)

AI Designer API is designed to run in containerised environments:

Docker Deployment

• Container Technology: Docker
• Application Stack: FastAPI + Gunicorn + Uvicorn
• Port Mapping: Map container port 8000 to the host

Resource Limits (Docker):

resources:
  requests:
    memory: 4Gi
  limits:
    memory: 6Gi

Kubernetes Deployment

Resource Configuration:

resources:
  requests:
    memory: 4Gi
  limits:
    memory: 6Gi

Operating System Requirements

Supported Platforms

• Preferred: Ubuntu Linux (latest LTS versions).

• Compatible: Any OS supporting Docker/Kubernetes.

Since deployment uses Docker/Kubernetes, the underlying OS is flexible. Container orchestration handles all application dependencies and environment setup.

Network Requirements

Port Configuration

Firewall Rules

• Allow incoming connections on port 8000 from Joget DX instance IP addresses.
• Allow outgoing HTTPS (443) for external AI service API calls.
• Configure firewall rules based on your network security policies.

Network Connectivity

• The Joget DX instance must be able to reach the AI Designer API on port 8000.
• Internet connectivity is required for communication between the container and external LLM services.

Dependencies

Application Dependencies

All dependencies are managed by Docker containers:

• Python runtime and libraries.
• FastAPI framework.
• Uvicorn ASGI server.
• AI/ML libraries and models.

Database Requirements

• No external database required.
• Application is stateless or uses internal data management.

Deployment Checklist

Pre-Deployment

• [ ] Verify server meets minimum specifications (4GB RAM, 4 cores).
• [ ] Install Docker or configure a Kubernetes cluster.
• [ ] Plan network architecture (separate vs same server).
• [ ] Configure firewall rules for port 8000.
• [ ] Ensure network connectivity between Joget DX and the Intelligence server.

Deployment

• [ ] Pull/build Joget Intelligence Docker image.
• [ ] Configure resource limits (4-6GB memory).
• [ ] Start container/pod and verify port 8000 is accessible.
• [ ] Test connectivity from the Joget DX instance.
• [ ] Install and configure the AI Designer plugin in Joget DX.

Scaling Considerations

Vertical Scaling

• Increase memory allocation up to 8GB for better performance
• Add CPU cores as needed based on concurrent request patterns
• Monitor resource utilisation and adjust accordingly

Horizontal Scaling

• Deploy multiple AI Designer API instances
• Use a load balancer in front of instances
• Configure the Joget DX AI Designer plugin with a load balancer endpoint

Additional Resources

• For detailed setup instructions, refer to Setting up AI Designer API Server.
• You can also download the deployment checklist PDF here:
  • Joget AI Designer API Server Deployment Checklist.pdf