AI Central Configuration plugin

Introduction

The AI Central Configuration plugin provides a unified interface to manage core configurations for AI services within a Joget application. It centralizes setup and control, ensuring smooth integration, testing, and governance of AI functionality.

Where to get the plugin

This plugin is bundled with the AI Agent Builder plugin.

How to use?

AI Central Configuration

AI Credentials

AI Credentials allow users to store API Keys from various AI providers as credentials, allowing for easy reusing and referencing when creating new AI Services (detailed in the next section).

1. Click Add Credentials.
   
2. Click Configuration Type and select an API key provider.
   
3. In step 2, specify the API key.
   
4. Give the credentials a name.
   

AI Services

Allows users to create, configure, and select AI services that can be called when using AI-powered plugins and features throughout Joget.

1. In Joget, click Settings and click AI Central Config.
2. In the AI Services tab, click Add AI Service.
   
3. In step 1, for Use Saved Credential (Optional), select AI credentials configured in the AI Credentials tab. Optionally, you can select None (Manual Setup).
4. For Model Class, select one of the following AI model classes from the drop-down list:
   
   
   • Anthropic
   • Google Gemini
   • OpenAI
   • Custom
5. In step 2, for Model Name, select the model version from the drop-down list.
6. Enter the LLM API Key for API Key.
   
7. In step 3, specify a name for the AI service.
   
8. Click Add.

AI Plugin

Click the AI Plugin tab. Here you can view and install all available Joget Intelligence plugins instantaneously without having to browse the Joget Marketplace.

AI Central Config Properties Configuration

AI Services

• Current AI Service:
  • Active AI Service: Select an AI Service to be used for generation from the dropdown.
• Manage AI Services:
  • Add AI Service: Create and customize your AI model to use.
    • Model Class:
      • Anthropic
      • Google Gemini
      • OpenAI
      • Custom
    • Custom LLM URL: Only required when selecting a Custom Model Class
    • Model Name: Select from a list of models available from the Model Class of choice. This field is named Custom Model Name if a custom model is selected.
    • API Key: Provide an API Key.
    • AI Service Name: Provide an identifiable name for the AI Service when using it within Joget.

AI Usage Dashboard

The AI Usage Dashboard is located in the Monitor tab, accessible from the Administrator Bar.

Here, you can also track the LLM calls made and tokens consumed through the AI usage dashboard. The AI usage dashboard consists of 3 sections:

• The main charts show the total prompt tokens and total completion tokens,
• The sub-charts show the prompt and completion tokens for each of the AI services,
• And the LLM calls table to keep track of the LLM calls made by the AI services

On the top of the dashboard, the two main charts show the Total Prompt Tokens chart on the left, and the Total Completion Tokens chart on the right as bar charts. The total consumed tokens will be shown on the top left of each respective chart.

In the Sub Charts area, for each AI service, there are Prompt Tokens and Completion Tokens charts for each model used. These charts are presented in the form of a bar chart, and the total tokens for each chart are displayed at the top left of the chart, labeled under their respective names. Their name changes depending on which tab is currently selected. Currently, there are two tabs available in the AI Usage Dashboard: The By Model Class and By Plugin Name tabs. These tabs can be used as sorting tabs.

By choosing a tab, the charts shown below will change accordingly based on the tab chosen; their naming format will also be changed. For example, by choosing the By Model Class tab, the charts will change their name based on the AI Model classes used, and the total tokens shown will be based on that specific Model class.

While for the By Plugin Name tab, the charts will be sorted based on the plugins that are using the tokens, and will show the total Prompt Tokens and Completion Tokens used by that plugin.

In the LLM Calls section, the Table shows and keeps track of all of the LLM calls done. The Table consists of the ID, Plugin Name, Model Name, Model Class, the creation date of the call, the Prompt tokens, and the Completion Tokens.

Skills

The Skills tab keeps the skills that ship with the product in sync and shows when an update is available. For what skills are and how agents use them, see Skills & Central Instructions.

Video Tutorial

The video below provides a visual representation of the steps mentioned.

Plugin Changelogs

• 8.2.0:
  • Added a new MCP Client Tool
  • New Agent Execution Audit Trail page available under the Monitor tab