How to access, monitor, and manage agent performance and tasks?
Last updated
Last updated
To ensure your agents operate smoothly and efficiently, it is essential to access, track, and manage their performance and tasks effectively. Here is a step-by-step guide on how to monitor task status, view detailed logs, filter and review tasks, and assess agent performance.
You can access all the agents you have created via the Agents Page. Here, you will find key details such as:
Agent ID: Unique identifier for each agent.
Agent name: The name of the agent.
Accuracy: The performance accuracy of the agent.
Action required: Any actions that need attention.
Tasks completed: The total number of tasks successfully processed by the agent.
Last task executed: The date and time the agent most recently completed a task.
Status: Whether the agent is active or in draft mode.
By reviewing these fields regularly, you can keep track of each agent's performance and ensure any necessary interventions are made promptly.
The reporting dashboard provides in-depth insights into an agent’s activity and performance. To access the dashboard for a specific ZBrain Agent, follow these steps:
Navigate to the agents tab: In the top navigation bar, click on the Agents tab to view the list of agents.
Select the agent: Find and select the agent you want to analyze. You can search for it by name or filter based on specific criteria.
Click on the agent name: This action will open the agent’s execution interface, where you will see several sections:
Queue: Displays tasks that the agent is currently processing or has completed.
Input: Shows the input data that the agent is processing.
Report: Provides detailed reports on task execution.
Activity: Displays the agent’s activity logs, including step-by-step execution details.
To add new tasks or documents for the agent to process, click the ‘Upload’ button in the Input section. This allows you to upload files directly from your system, enabling the agent to begin processing them.
You can view the queued or processed documents by searching through the documents in the relevant sections of the dashboard. The system will allow you to filter and view documents based on their status.
Detailed processing reports are available for review directly from the Report section. These reports can be downloaded or copied to review results or share them with team members for further evaluation.
If the results of a task do not meet your expectations, you can provide feedback directly through the dashboard to help improve the agent’s performance over time.
Thumbs down button: If a task does not meet the desired standards or fails to produce accurate results, you can give a thumbs down to indicate dissatisfaction. This feedback helps refine and optimize the agent’s performance for future tasks.
To gain deeper insight into how the agent is processing tasks, you can view the session logs by expanding the Agent Activity section. This will provide you with detailed information about each step in the agent’s execution process:
Input data: Understand what data the agent has processed, including the format and type of documents it is handling.
Output data: See what results or outputs the agent has generated after processing the input.
Step-by-step overview: A breakdown of each step during task execution, including:
Status of the step: Whether the step was successfully completed, failed, or skipped.
Input: Data the agent received at each step.
Time taken: The total duration the agent took to complete a specific step.
Output: What the agent produced after each processing step.
Reviewing these logs allows you to see exactly how the agent processes tasks, helping you identify areas for improvement or optimization.
On the Agent Dashboard, you will see indicators that represent the status of each task in real-time. These indicators help you quickly assess whether your agent’s tasks are progressing as expected:
Green dot: Indicates that the task has been successfully completed without any issues.
Yellow dot: Signifies that the task is pending and awaiting processing. This may happen if there’s a queue of tasks or a delay in processing.
Red dot: Shows that the task has failed, indicating that there were issues with processing, such as errors or system failures.
Monitoring task statuses in real-time helps ensure you are aware of any issues and can take corrective action promptly.
Managing the document pipeline effectively is crucial for optimizing agent performance. Use the filter and sort tasks by their status:
Show all: Displays all tasks regardless of their status.
Processing: Tasks that are actively being executed by the agent.
Pending: Tasks that are waiting for the agent to start processing.
Completed: Tasks that have been successfully finished by the agent.
Failed: Tasks that encountered errors and were not completed successfully.
You can apply filters by clicking on the filter icon, and the filtered tasks will appear on the right side of the screen for quick review and management. This feature makes it easier to track the progress of specific tasks and manage your workflow efficiently.
The Performance section of the agent’s dashboard provides detailed metrics on how well the agent is performing over a specific period:
Utilized time: The total amount of time the agent has been actively processing tasks.
Average session time: This metric shows the average duration it takes for the agent to process a task.
Satisfaction score: An average score reflecting the quality of the tasks performed by the agent. This score could be based on user feedback.
Token used: The total number of tokens consumed by the agent during its tasks. This also provides an estimate of the associated processing costs.
A detailed breakdown of each task processed by the agent is provided. Key columns include:
SessionID: Unique identifier for the processing session.
Record name: The name or identifier for the specific task processed.
Session start date: When the processing of the task began.
Session end date: When the processing concluded.
Session time: The total duration of the task.
Satisfaction score: The rating assigned to the task.
Token used: Tokens consumed during the session.
Cost: Estimated cost associated with processing the task.
This information helps you assess agent performance at a granular level.
The All Flows page allows you to evaluate the specific workflows (flows) and triggers used by the agent during task execution:
Flow name: Each workflow the agent uses is identified by name.
Flow status: Displays the current status of the flow, such as active or draft.
Updated: Shows the last date and time the flow was modified.
Reviewing flows helps you understand how the agent’s processes are structured and whether they are optimized for efficiency.
By regularly accessing, monitoring, and managing your agent's tasks, performance, and flows, you can ensure smooth operations, optimize efficiency, and make data-driven adjustments for continuous improvement. This proactive approach helps maintain high performance and enables seamless workflow automation.
