Creating Agent

To begin creating an agent, click on the Create Agent button and follow the guided setup flow.

The Create an Agent window opens. The left‑hand panel displays guided steps to help you move through the setup process, while the right‑hand panel displays the Agent Basic Setup fields that you must complete.

Agent Basic Setup

• Name: Enter a meaningful name for the agent. The name helps identify the agent’s purpose and distinguishes it from other agents in the organization.
  Must start with alphanumeric, with no space, contain only alphanumeric, hyphens (-) or underscores (_), but no consecutive hyphens (-) or underscores (_)

• Goal & Context: Provide clear, outcome‑focused instruction that explains who the agent is and what it aims to achieve. This definition guides the agent’s behavior during execution.

  The instruction must be at least 40 characters long. Use concise and specific language to clearly describe the agent’s intent.

• Guardrails: Define the rules and boundaries within which the agent must operate. Guardrails describe what the agent must not do, should avoid, or must stay within while achieving its goal. This ensures the agent behaves safely, ethically, and appropriately.

• Base Models: Agents are powered by a Large Language Model (LLM) that enables reasoning, planning, and decision‑making capabilities.

  Select the model used for the agent. By default, the agent uses Claude. The following models are available from the dropdown list:

  • Claude 3.5 Sonnet

  • Claude 4 Sonnet

  • Claude 4.5 Sonnet

When you have completed the required information, select Next to continue with the agent setup, or select Cancel to exit and discard the agent creation process.

 All fields in this step are required and must be completed to proceed with agent creation.

Knowledge

Agent knowledge defines the information an agent has access to when it starts executing. This can include background context, documents, company data, FAQs, or other reference material. The agent uses this knowledge to reason more effectively and produce relevant decisions or responses.

After the Agent Basic Setup details are validated, the setup flow moves to the Knowledge page.

Goal and Context

Provide the goal and context for the knowledge being added. This helps the agent interpret and apply the uploaded information while performing its tasks.

Upload Knowledge Files

You can add knowledge by uploading a file using one of the following methods:

• Upload the file using the file picker, or

• Drag and drop the file into the upload area

The following behavior applies when uploading a file:

• Supported file types include PDF or XML

• The maximum file size is 20 MB

Once uploaded, the file details (File type (PDF or XML), File name, File size)  are displayed in the right-hand section. Also, the Next button changes to Process Files.

Select Process Files to process the uploaded file. After the file is successfully processed, the Process Files button changes to Next. Select Next to continue with the agent creation process.

Managing Uploaded Files

• Hover over an uploaded file to display the Delete icon, which allows you to remove the file.
  

• After a file is successfully processed, the delete icon is removed, and a green success indicator is displayed next to the file.
  

• A processed file can be replaced by dragging and dropping a new file into the upload area. The new file automatically replaces the existing one.

Tools

Tools are the actions an agent can take to interact with the world, such as calling an API, searching the web, running code, querying a database, or sending a message. They extend the agent’s abilities beyond just thinking and talking.

Xponent Graphs

In this step, you choose which Xponent Graphs the agent can use as callable tools. If no Xponent Graphs are available for agent use, the right‑hand section displays the message “No Xponent Graphs available. Publish Graphs for Agent use in your organization.”

Just below the Xponent Graphs section, the UI displays how many tools are available, for example: 02 tools available. This indicates the maximum number of Xponent Graphs that can be selected as tools for the agent.

Once graphs are available, you can select Xponent Graphs from your projects to allow the agent to invoke them as tools. Start by selecting an environment. The available environments are:

• Production

• Pre‑Production

• Non‑Production

After selecting an environment, the list of graphs deployed in that environment is displayed along with their deployment status.

• Graphs can be searched using the search (Find) box.

• You can select multiple graphs within the same environment.

• If you change the selected environment, any previously selected graphs are cleared.

• For Non‑Production environments, an additional Environment column is displayed to indicate whether the graph belongs to QA, Development, or another environment.

• Once a ßgraph is selected, graph names appear in the left‑hand panel along with the environment they are associated with. This allows you to review which graphs the agent can access before continuing.

API Calling

The API Calling feature allows your agent to call external APIs during execution. This enables the agent to retrieve data from external systems or trigger actions in downstream services, depending on the API's capabilities. In the setup panel, you define the API endpoint, authentication, and request details so the agent can call the API consistently and safely.

To create a new API tool, open the API Calling setup panel and provide the endpoint details you want the agent to call. Each tool represents one API configuration that the agent can later select and use.

Tool Details

• Name: Enter a clear, descriptive name that makes it easy to recognize the purpose of the tool. For example, use a name that reflects the outcome, such as Get Customer Profile or Create Support Ticket, rather than a generic name like API 1.

• Description: Add a short explanation of what the API does and when the agent should use it. A good description helps other configurators understand the intent and reduces misuse. For example, you can mention what the API returns or what action it performs.

• URL: Enter the complete API endpoint URL. This should include the base path and resource path required by the API. If the endpoint requires dynamic values (such as an ID), you can use variables in the URL (see Variables section below).

Request Method

Select the HTTP method the agent should use when calling the API. The supported methods include:

• GET - Retrieve information (read-only requests).

• POST - Create or submit information (commonly used when sending a body).

• PUT - Replace an existing resource.

• PATCH - Update part of an existing resource.

• DELETE - Remove a resource.

Choosing the correct method is important because many APIs validate the method and may reject requests that do not match the expected operation.

Request Configuration

• Headers (JSON): Use headers to provide additional request metadata, such as content type, accepted response format, correlation IDs, or API keys (when applicable). Headers must be provided in JSON format.

• Params: Use params (query parameters) to pass values in the URL query string. This is common for filters, pagination, or search inputs. If your UI expects params as key/value pairs, add them accordingly. If it expects JSON, use a JSON structure similar to headers.

• Request Body (JSON): Use the request body to send data for methods such as POST, PUT, or PATCH. The body must be in JSON

Authentication

Authentication determines how the agent proves it is allowed to call the endpoint. Select the authentication type that matches the API requirements:

• None: Use this only when the endpoint does not require authentication.

• Basic Auth: Use Basic Auth when the API requires a username and password. Provide:

  • Username

  • Password

• API Key: Use API Key authentication when the API expects a key passed in a header. Provide:

  • Header Name (for example, x-api-key or Authorization)

  • API Key Value

Variables

Variables allow you to insert dynamic values into the tool configuration so the agent can call the same API with different values at runtime. You can use variables in the URL, params, headers, or request body, depending on the supported fields.

• To create a variable, type it in the following format: {{variableName}}

After you enter the required details:

• Select Create Tool to validate and save the tool configuration.

• The button shows Validating while the tool is being checked and created.

• Once validation succeeds, the tool becomes available in the tools list and can be selected by the agent for use.

• When selected, the tool appears in the left-hand panel, confirming it has been added to the agent configuration.

You can update a created API tool at any time:

• Locate the tool in the tools list and select the Edit (pencil) icon.

• Update the fields you want to change (URL, method, auth, headers, params, or body).

• Select Update Tool to save changes.

This allows you to refine the tool configuration without recreating it from scratch.

Generic Tools

In addition to Xponent Graphs and API tools, you can enhance an agent’s capabilities by adding Generic Tools. Generic tools provide built‑in functionality that agents can use without additional configuration.

To add a generic tool:

• Select the Add a tool option.

• A list of available generic tools is displayed.

• Select the tool you want to add. For example, select Web Search.

The Web Search tool allows the agent to search publicly available information on the web to support its reasoning or decision‑making. This is useful when the agent needs additional context or up‑to‑date information that is not part of its configured knowledge base.

Once selected, the Web Search tool is added to the agent’s tool list.

Review Selected Tools

All selected tools, including generic tools, appear in the left‑hand panel, allowing you to review which tools the agent has access to before completing the setup.

After selecting the required tools:

• Review the agent configuration to ensure all required steps are complete.

• Select Create Agent to complete the agent creation.

Confirmation

After you select Create Agent, a confirmation message is displayed indicating that the agent build process has started.

The system displays the following confirmation:

Agent build started successfully.

Your agent is now being prepared. This Agent ID is shown only once. When you close this window, it will be copied automatically. Paste it into the Agent ID field to link your agent.

Setup usually takes a few minutes - you will receive a notification when the agent is ready.

Agent Build Details

The confirmation window displays the following details for reference:

• Agent Name

• Model used by the agent

• Tools selected for the agent

• Build Started On timestamp (For example: Tue Apr 21 2026 20:24:06 GMT+0000 (Coordinated Universal Time))

Copy Agent ID and Close

• Select Copy Agent ID & Close to copy the Agent ID and close the confirmation window.

• Also, the Agent ID is copied automatically when the window is closed.

Once the build process starts, the agent status is set to Not Ready. After the build process completes successfully, the agent status updates to Ready.

 If the agent build does not complete successfully, the agent remains in the list with a Not Ready status. The agent is not removed automatically.

Agent readiness notification

Once an agent is created and successfully provisioned, the system displays a notification confirming that the agent is ready to use. For Example, “Agent Profile_Data (c8752cfd9328e8d5) is now ready to use.”

The notification typically appears within 20–30 seconds and is shown only if the page is not refreshed during this time. If you do not receive the notification, refresh the page. The agent status will update accordingly.

If the agent remains in Not Ready status, raise a support ticket for investigation and create a new agent if required. Agent deletion is supported only through a support ticket.

 Agent can be used in Journey Orchestration only through Managed Graphs. A native Agent node is not available in the node palette. To get started and build an agent for your use case, contact an Xponent Consultant.