AI Agent analytics

Applies to version: 2026 R1 and above; author: Adrian Baszak, Lily Adamowicz

Related documentation

A detailed description of the functionalities mentioned herein and their configuration can be found in the following sections of the WEBCON Help:

• AI Agent

• AI Implementation Guide

Other related Community articles include:

• AI Agents – Business use case

Introduction

The AI Agent analytics dashboard is a centralized monitoring and audit interface designed for administrators responsible for overseeing AI-driven processes. It provides a consolidated view of AI Agent activity and historical execution data, enabling both real-time supervision and post hoc analysis of AI Agent behavior within the system. 

Dashboard overview 

The dashboard consists of several clearly defined components, each addressing a different aspect of AI Agent operations: 

Fig.1. AI Agent analytics dashboard 

1. Active Agents—a widget displaying the number of currently active AI Agents in relation to the total number of AI Agents configured globally. 
2. Queue depth—a metric representing the number of jobs currently waiting in the processing queue. 
3. Success rate—a widget presenting the percentage of successfully completed jobs relative to all processed jobs. This metric allows administrators to quickly identify degradation in AI Agent performance. 
4. Agent fleet—a list of all configured AI Agents, grouped by the processes to which they belong. For each AI Agent, the report also displays the associated step, a workflow, and the current status. This supports governance and lifecycle management of AI Agent configurations. 
5. Live job queue—a real-time list of jobs currently being processed or awaiting execution. Each entry includes context as well as the job waiting time, allowing administrators to identify long-running tasks and react promptly. 
6. Execution history 

• Administrators can use the Inspect action button to check detailed execution data, including the AI Agent’s decision logic and processing context: 

Fig.2. AI Processing Inspector 

Technical architecture 

The solution is implemented as a lightweight, client-side visualization layer that operates directly on top of WEBCON’s native platform infrastructure. It is not a built-in dashboard, but a configurable extension that requires explicit setup within the platform environment. All operational data—including AI Agent definitions, queue states, and execution history—remains stored within WEBCON’s internal database. 

Data Extraction and security 

Data is retrieved via User Defined APIs, which are custom API endpoints configured within the platform. These endpoints execute predefined SQL queries and expose selected internal system tables as structured JSON responses. 

Three User Defined APIs were prepared for the purpose of providing data displayed on the dashboard: 

• AI Agents—returns a list of all AI Agents defined in all workflows, 
• AI Agents Jobs—returns a list of all current jobs in progress or scheduled to be executed, including workflow instance ID and planned date of execution, 
• AI Agents Executions—returns a list of all completed jobs, including workflow instance ID, date of execution and AI Agent reasoning. 

Key characteristics of this approach include: 

• Real-time data access—the dashboard consumes data directly from the source, ensuring accuracy and up-to-date information. 
• Secure execution context—the dashboard runs entirely in the user’s browser and leverages existing session cookies and platform authentication mechanisms to access API endpoints securely. 

For more details on configuring and using User Defined APIs, refer to the dedicated documentation article.