Execution Details

The Execution Details page provides a complete step-by-step breakdown of how an AI Colleague processed a specific request—from initiation to completion. Access this view by clicking any row in the Executions Table.

Key Components:

Header: Run metadata including run ID, status, timestamps, and token usage
Run Execution Timeline: Vertical stepper showing each step in chronological order
Section Details: Detailed information for the selected timeline step
Side Panel: Expanded details for specific section types

Header

The header bar displays key information about the run at a glance.

Element	Description
Initiated On	Date and time when the run started
Completed On	Date and time when the run finished
Status	Outcome: Handled (green), Unhandled (red), In Progress (blue), Cancelled (orange)
Total Time Taken	Duration from start to finish
Input Tokens	LLM tokens consumed as input
Output Tokens	LLM tokens generated as output
AOPs/Skills Utilized	Click "View List" to see all AOPs and Skills used during execution
Initiation Details	Click "View Details" to see who initiated, trigger type, and channelInitiation Details

Run Execution Timeline

A vertical stepper showing each step the AI Colleague took to process the request.

Step Components

Component	Description
Icon	Visual indicator of step type
Tittle	Name of the action (e.g., "Run initiated", "Eligible AOPs/Skills")
Initiated On	Timestamp when this step started
Status Chip	Green = Success, Blue = Pending, Red = Failed
Nested Steps	Some steps expand to show sub-steps (e.g., AOP with multiple actions)

Step Types

Step Type	Description
Run Initiated	User's initial message was received
AI Reasoning	LLM analysed request and decided what action to take
Knowledge Search	Searched knowledge base for relevant information
Skill Execution	Called an API or ran a workflow skill
AOP Execution	Started an Agent Operating Procedure (multi-step workflow)
Form Collection	Collected information from user via conversation
Loop Execution	Processed multiple items in a loop
Approval Request	Sent request for human approval
Escalation	Escalated to human agent

Section Details

When you click a timeline step, the right panel shows detailed information about that step. The content varies based on the step type.

Section Types

Section	Description of Data
Table Section	Data in rows/columns format
Chat Threads	Conversation between user and bot
Form Details	Collected form field values
Request/Response	Raw JSON sent to/received from API
AI Decision Summary	LLM reasoning for action selection
Loop Execution	Table of loop iterations
Approval Details	Approvers, status, timestamps
View Action Details	Link to drill into nested workflow

Side Panel

Clicking certain section cards opens a side panel with expanded details.

Panel Type	Content
Chat Threads	Full conversation with user messages and bot responses, timestamps, and option to view raw request/response for any message
Form Details	All form fields with labels, collected values, and validation status
Request/Response	Two JSON code editors: one for request payload, one for response. Includes prettify option
Knowledge Search	List of retrieved articles with name, type, source, and relevant excerpts used

Troubleshooting Guide

Issue	Solution
Find why a run failed	Look for red status chip in timeline, click that step, check Request/Response for errors
See what user said	Click "Chat Threads" section to see full conversation
Debug an API call	Click skill execution step → Request/Response Payload → view JSON
Check token usage	Look at Input/Output Tokens in header
See what AOPs ran	Click "View List" next to AOPs/Skills Utilised in header
View AOP internal steps	Click "View Details" on AOP step → use breadcrumbs to drill in

AOP Execution Details

View the step-by-step execution of Agent Operating Procedures

Accessing AOP Execution Details

To view AOP execution details:

Open any run from the Run History page
Find an AOP Execution step in the timeline
Click the step to see its details in the right panel
Click the "View Details" button to open the AOP execution timeline

The AOP Execution Details page has the same layout as the main run details page, with content specific to the AOP workflow.

Navigating Nested AOPs

When a main AOP calls a helper AOP, you can drill into the helper to see its execution details. This creates a navigation hierarchy that you can traverse using breadcrumbs.

Example:

A "Create Jira Ticket" AOP calls a "Sync to Spreadsheet" helper AOP. The breadcrumb trail would show:

Run Execution Timeline > Create Jira Ticket > Sync to Spreadsheet

Click "Create Jira Ticket" in the breadcrumb to go back to the main AOP view, or click "Run Execution Timeline" to return to the original run details.

Capability Matching

When viewing the execution details for any run, the Capability Matching section provides transparency into how the AI Colleague identified and selected the appropriate skills and AOPs to handle a user's request.

What is Capability Matching?

Capability Matching (also called Re-ranking) is the process by which an AI Colleague evaluates all available skills and AOPs against an incoming user request to determine which capabilities are most suitable for execution. This involves three stages:

Eligibility Assessment-Identifying all potential capabilities that could handle the request
Relevance Scoring-Ranking capabilities based on how closely they match the request
Final Selection-Choosing the optimal set of tools to present to the LLM for execution

Eligible AOPs/Skills

This table shows all capabilities initially considered for the request

Column	Description
Name	Display name of the skill or AOP
Type	Classification: Skill (Default/Custom) or AOP (with AI Colleague identifier)
Action	Click to view the skill or AOP configuration

LLM: AOP/Skill Selection

This table shows the final set of capabilities sent to the LLM and is similar to Eligible AOPs/Skills Table.