Agents

The Agent Node enables the use of AI‑powered, goal‑driven agents within Xponent workflows. It allows agents to be executed as part of graph orchestration, where they can apply goals, knowledge sources, and available tools to make decisions or perform actions within a journey. By using Agent Nodes, teams can reduce manual or repetitive workflow logic and introduce intelligent decision support directly into graphs.

Prerequisite

• Xponent Intelligence: An administrator must enable Xponent Intelligence at the organization level. Also, in the Xponent Intelligence settings, the Agent Node toggle must be enabled for you to create agents.
  When the Agent Node feature is turned off at the organization level, you can view existing agents but cannot create new ones. Any Agent Nodes already used in workflows continue to run without interruption.

When the Agent Node feature is turned on but no AI Agents exist in the organization, the Agent page displays an empty state to guide you in creating your first agent.

The empty state provides a brief explanation of the agent’s purpose and highlights how they can be used in Xponent workflows. A Create Agent button is prominently displayed to start the agent creation process.

 Until at least one agent is created, the agent list table, filters, and related actions remain hidden.

Sorting, Filtering, and Searching

The Agents page displays all AI Agents available at the organization level, allowing you to review their configuration, readiness, and access to Xponent Graphs before creating or managing agents.

Sorting

When one or more agents exist, the page displays a searchable and sortable list of agents. Each column provides key information at a glance, including:

• Agent Name: The name assigned to the agent during creation. This helps identify the agent’s purpose within the organization.

• Version: Displays the graph associated with the agent. Agents that are not currently used in any graph display

• Goal and Context: A brief summary of the agent’s defined goal and contextual information. If the text is longer than the column width, it is truncated. Hover over the Read more text to view the complete goal and context.

• Graph Environment: Indicates if the agent has access to Xponent Graphs and identifies the single environment it is configured for. For Example, Production, Pre‑Production, or Non‑Production

• Last Updated: Displays the date and time when the agent was most recently modified. For example: 12 Jun 2026 01:16.

• Updated By:

• Updated By: Displays the name of the user who made the most recent changes to the agent configuration.

• Status / Readiness:  Reflects the agent’s operational state based on internal readiness checks and indicates whether the agent is available for use in workflows or failed. For Example, 🟢 Ready or ⚠️ Failed.

• Actions

  • View: View the agent configuration, including its goal, context, knowledge sources, tools, guardrails, and status.

  • Edit: Modify the agent configuration, such as instructions, context, knowledge sources, and tools.

  • Delete: Remove the agent from Agent Builder if it is no longer required.

You can sort the list by Agent Name, Graph Environment, or Last Updated by selecting the column headers in ascending or descending order.

Filtering

The agent list provides quick filtering options to help you refine the agent list. You can select multiple filters simultaneously to further refine the agent list. For example, you can combine readiness filters with Graph Tools filters to quickly locate agents that match specific criteria.

Readiness Filters

• 🔶  Preparing: The agent is currently being configured or processed and is not yet ready for use.

• ⚠️ Failed:  The agent has encountered a configuration or processing issue and cannot be used until the issue is resolved.

• 🟢 Ready:  The agents that are operational and available for use in workflows.

Graph Tools Filters

• No Graph: Displays agents that do not have any graph tools associated with them.

• Production: Displays agents that are associated with Production graph tools.

• Non-Production: Displays agents that are associated with Non-Production graph tools.

These buttons display the count of agents matching that status. Selecting a filter updates the agent list immediately. Selecting the same filter again (double‑selecting) clears the filter and returns the list to its default state.

Searching

The agent list includes a search field that lets you search for agents by Agent Name, Agent ID, Goal, or Context. Once you enter this metadata, the list updates to show matching agents.

If no agents match the entered text, the following message is displayed: “No results found for entered text. Check for typos or try different keywords. You can also create a new agent if none match what you're looking for.”

This helps you refine your search or quickly create a new agent when needed.

 Status filters, search, and column sorting can be used together to narrow down large agent lists and quickly find the required agents.

Viewing Agent Configuration

You can view the configuration details of an existing agent directly from the Agents list.

• In the Action column of the Agents list, select the View (eye) icon for the required agent.

• The agent configuration opens in view‑only mode.

In view mode, the following sections are displayed for reference:

• Basic Setup: Displays the agent’s name, goal, and context, guardrails, and selected base model.

• Knowledge: Shows the knowledge configuration for the agent, including goal and context for the knowledge and any uploaded or processed knowledge files.

• Tools: Displays the tools configured for the agent, such as Xponent Graphs, API tools, or generic tools (for example, Web Search).

• Guardrails: Displays the guardrail configuration applied to the agent to help control agent behavior, content handling, and data protection settings.

 The agent configuration is read‑only and cannot be modified from the view screen.

Deleting Agent Configuration

You can permanently remove an existing agent and its associated configuration directly from the Agents list.

Before deleting an agent, verify whether it is currently used in any orchestration graph. Deleting an agent permanently removes its configuration and cannot be reversed.

To delete an agent:

1. Navigate to the Agents list and locate the agent you want to remove.

2. In the Actions column, click the Delete (bin) icon corresponding to the agent.

   

3. A confirmation dialog appears displaying the selected agent name and a warning message indicating that the deletion is permanent and cannot be undone.

4. Review the agent name carefully and verify that you are deleting the correct agent. If you do not want to proceed, click Cancel to close the confirmation dialog without deleting the agent.

   

5. Click Delete to proceed with the deletion. The system removes the agent and all associated configuration details from Agent Builder.

   Once the deletion is completed successfully, a confirmation message appears in the notification panel indicating that the agent has been deleted.

   

Editing Agent Configuration

You can edit the configuration details of an existing agent directly from the Agents list.

The agent name cannot be changed after the agent is created. All other agent configuration can be updated as needed.

To edit an agent:

1. From the Agents list, locate the agent you want to update. In the Action column, select the Edit (Pencil) icon.

   

2. Modify the required configuration, such as the Basic Setup, Knowledge Base, Tools, and Guardrails, and other configuration details.

   

Reverting Unsaved Changes

If you want to discard your current edits and restore the latest saved version of the agent configuration:

1. Select the More Actions (Vertical Ellipsis⋮) menu and click Get Latest Configuration.

2. In the confirmation dialog, before proceeding, read the message: Are you sure you want to revert to the latest configuration? All current changes will be lost.

3. Click Revert to restore the latest saved configuration, or click Cancel to continue editing.

   

Reviewing Configuration Changes

After completing your updates, select Save & Apply located in the upper-left corner of the page.

The Review Changes page displays a summary of all configuration changes:

• Settings: The configuration area where changes were made, such as Tools or Guardrails.

• Item: The name of the configuration item that was modified.

• Current Version: Previously saved details. This information is typically displayed for removed or modified changes.

• New Version: Updated details. This information is displayed for newly added or modified content.

• Change: Indicates the type of change, such as Added, Removed, or Modified.

For modified configuration items, the review page displays both the previous and updated values. The Current Version column shows the older details, while the New Version column shows the newly added details that will be applied when you save the changes.

To view configuration items that have not been changed, enable the Show Unchanged toggle. Reviewing unchanged items can help you verify that no additional updates are required before saving.

Saving and Applying Changes

Saving creates a new current version of the agent. Any workflows, automations, or integrations that use the agent will automatically run with the updated version.

To save your updates:

• Review the configuration changes.

• Click Save & Apply. In the confirmation dialog, confirm that you want to apply the changes.

A success message appears indicating that the updated agent configuration is now live in the production environment. After the changes are successfully applied, the updated agent is displayed in the Agents list.

Best Practices

• Give each agent a focused responsibility. Agents with a single, well‑defined goal are easier to understand, test, and reuse across multiple graphs.

• Include only the information the agent needs to perform its task effectively. Excessive or unrelated context can reduce clarity and impact decision‑making.

• Avoid creating multiple agents with overlapping goals. Reusing agents helps maintain consistency and reduces configuration overhead at the organization level.

• Grant agents access only to the tools and Xponent Graphs required for their purpose. This helps minimize unintended behavior and simplifies troubleshooting.

• Assign clear names and meaningful goal descriptions so other users can easily understand what the agent is designed to do.

• Begin with a basic configuration and expand the agent’s knowledge or tools as requirements evolve. This makes testing and validation easier.

FAQ