Agent Work Tracking

Simple Agent Designer tracks all work performed by AI agents, providing visibility into what your agents have been doing. This page explains the Agent Work tracking system.

Overview

Every time an agent completes a significant action — running a report, creating a journal entry, or processing a workload — it is recorded in the Agent Work tracking system. This provides:

A summary of recent agent activity visible on the Role Center
An audit trail of AI-initiated actions
Quick access to generated outputs (reports, documents)
Per-user work tracking

The Agent Work Part

The Agent Work part appears on the Agent Based Accounting Role Center. It displays a list of recent work entries, giving you an at-a-glance view of agent activity.

What is Tracked

Agent Work entries are created for various types of agent actions:

Work Type	Description	Example
Agent	General agent processing completed	Workload analysis completed
Report	A Business Central report was generated	Sales report for Q1 generated
Data	Agent created or worked with data records	GL journal entry for invoice INV-001

There is no separate "Journal" work type. Journal entries created by agents are surfaced under the Data type; when the underlying table is a General Journal Line, the entry's drill-down opens the General Journal directly.

Work Entry Fields

Each Agent Work entry contains:

Field	Description
User	The Business Central user who initiated the processing (UserSecurityId)
Type	The category of work (Agent, Report, Data)
Name	A descriptive name for the work performed
Language	The user's language setting
Last Action	The most recent action or status update
TableNo	The number of the Business Central table the work relates to (for Data entries)
Table Name	The caption of that table, for readability
Output Format	The format of the generated output: PDF, XML, HTML, Markdown, or Excel
Output Binary	Generated file content (e.g., report PDF or generated Excel file) — stored as a blob
Output Name	Filename of the generated output
Workload	Reference to the associated workload (GUID)

Viewing Work Details

Click on a work entry to see its details
If the entry has an Output Binary (e.g., a generated report or an Excel file), you can download it directly from the list
For Data entries backed by a Gen. Journal Line, drill down to open the General Journal
Use the Workload reference to navigate to the full workload card and process history

How Work Entries Are Created

During Workload Processing

When a workload is processed in the background (via the Background Process codeunit):

The orchestration engine processes the workload
Upon completion, an Agent Work entry is created
The entry links to the processed workload

During Interactive Agent Use

When you use the Interactive Agent on the Role Center:

The agent performs actions (tool calls, report generation)
Relevant actions are tracked as Agent Work entries
Entries appear in the Agent Work part

During Report Generation

When an agent uses the run_report tool:

The report is executed with the specified parameters
The output is stored in the Agent Work entry's Output Binary field
The Output Name records the filename

By User

Agent Work entries are associated with the user who initiated the processing. The Role Center typically shows entries for the current user.

By Type

You can filter by work type to see only specific categories:

Agent — general processing activities
Report — generated reports
Data — data records the agent created or worked with, including journal entries

Drill-Down to Workloads

From any Agent Work entry with a Workload reference, you can navigate to:

The Workload Card — view the full task description and results
The Process Activity — see the detailed step-by-step execution log
The Documents Created — see all business documents linked to the workload

Using Agent Work for Auditing

The Agent Work system provides an audit trail for AI-initiated actions:

Who started it: The User field identifies who initiated the processing
What was done: The Type and Name fields describe the action
When it happened: The Last Action field shows the most recent activity
What was produced: Output Binary contains any generated files
Full context: The Workload reference links to the complete processing history

For compliance purposes, combine Agent Work entries with Process Activity records for a complete picture of every step the AI took.

Tips and Best Practices

Review the Agent Work part regularly to stay informed about agent activity
Download and review generated reports before distributing them
Use the Workload links to investigate unexpected agent actions
If you need detailed step-by-step auditing, always check the Process Activity records
Agent Work entries accumulate over time — consider periodic review and cleanup for large volumes