Skip to main content

​
Understanding the Agent Creation UI

The Agent Creation feature is integrated into the platform’s workflow configuration settings. The UI is organized into several key sections to help you create, configure tools for, and deploy Agents. Below, I’ll describe the UI for each major section and explain how to interact with it.

​
1. App Configuration UI

Purpose: Manage Agents participating in an App’s workflow. How to Access:
  • Navigate to the App where you want to add or manage Agents.
  • Click on the “Workflow” tab or a similar option in the App settings menu.
UI Elements:
  • Agents Section: Displays a list or table of existing Agents assigned to the App, showing the Agent’s name and a brief description.
  • “Create Agent” Button: Initiates the process of creating a new Agent, located near the Agents list.
User Interaction:
  • Scroll through the Agents section to view existing Agents.
  • Click the “Create Agent” button to open the Agent Configuration form.
Tip: If no Agents are listed, a message like “No Agents assigned yet” with a prompt to create one may appear.

​
2. Agent Configuration UI

Purpose: Create and configure a new Agent, defining its name, model, and behavior. How to Access:
  • From the App Configuration UI, click the “Create Agent” button.
UI Elements:
  • Form Layout: A structured form with labeled fields and buttons.
  • “Name” Field: Text box for a unique Agent name (e.g., “SupportBot”).
  • “Description” Text Area: Larger text box for the Agent’s purpose (e.g., “Handles customer support queries”).
  • “LLM Model” Dropdown: Lists available models (e.g., Claude Sonnet 3.5). Displays “No models available” if none are configured.
  • “Set up AI Service” Link/Button: Appears next to the LLM dropdown if no models are available, linking to AI Service setup.
  • “Training Instructions” Text Area: Field for behavior guidance (e.g., “Ask users for issue details before creating a ticket”).
  • “Start Conversation” Button: Opens a chat interface to test the Agent.
  • Chat Interface: Pop-up or inline window with a text input for messages and a response display area.
  • “Clear Conversation” Option: Button or link in the chat interface to reset the conversation.
  • “Save” Button: Saves the configuration and redirects to the Agent’s profile page.
  • Agent Profile Page: Displays the saved configuration with an “Edit” button to modify settings and a “Start Conversation” button for testing.
User Interaction:
  1. Enter a name and description in the respective fields.
  2. Select an LLM model from the dropdown; if none, click “Set up AI Service” to configure one.
  3. Add instructions in the “Training Instructions” text area.
  4. Click “Start Conversation” to test (e.g., type “Help me with a login issue”), and use “Clear Conversation” to restart if needed.
  5. Click “Save” to create the Agent and view its profile.
  6. On the profile page, use “Edit” to adjust settings or “Start Conversation” to test further.
Example: For L1 support, name it “SupportBot,” describe it as “Handles initial customer queries,” select Claude Sonnet 3.5, and instruct it to “Ask for issue details and priority.”

​
3. Tools Configuration UI

Purpose: Add and configure tools for Agents to perform actions like creating or searching records, running automations, delegating to sub-agents, or connecting to external MCP services. How to Access:
  • From the Agent Configuration UI or Agent Profile Page, locate the tools section and click the “Add Tool” button or menu.
UI Elements (General):
  • “Add Tool” Button/Dropdown: Opens a menu with tool types (Search, Search Records, AI Search, Create Record, Update Record, Run Automation, Run Agent, MCP Tool).
  • Tool Configuration Form: Appears after selecting a tool, with specific fields.
  • “Tool Name” Field: Text box for naming the tool.
  • “Description” Text Area: Field for usage guidance.
  • “App/Element/Task” Dropdowns: Menus to select the applicable App, Element, or Task.
  • “Save Tool” Button: Saves the tool configuration.
Specific Tool UIs:

​
Create Record Tool

  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions:
    • Run as publisher (default): Agent runs with publisher’s permissions.
    • Run as current user: Agent runs with current user’s permissions.
    • Run as service account: Agent runs with a specific service account.
  • “Fields” Section: Shows required fields by default with an “Add Optional Field” button.
  • Field Description Text Boxes: Guidance next to each field (e.g., “Ask the user for a brief summary”).
  • “Return Fields” Checkboxes: List of fields to return to the Agent after creation.

​
Record Search Tool

  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions (same options as Create Record Tool).
  • “Query Description” Text Area: Guidance on what to search (e.g., “Search for tickets by issue type”).
  • “Return Fields” Section: Optional checkboxes for fields to return.

​
Update Record Tool

  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions (same options as Create Record Tool).
  • “Record Identification” Text Area: Explains how to identify the record (e.g., “Use the ticket ID provided by the user”).
  • “Fields to Update” Section: Select fields to update, mark as required if needed, and add descriptions.

​
AI Search Records Tool

  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions (same options as Create Record Tool).
  • “Result Limit” Slider/Input: Sets maximum results (1-100, default 10).
  • “AI Search Configuration” Dropdown: Selects the AI Search setup.
  • Attribute Fields and Filters: Options to choose searchable fields and filters with description text boxes.
  • “Return Fields” Checkboxes: Selects fields to return from search results.

​
Run Automation Tool

  • “Tool Name” Field: Text box for naming the tool (no spaces or special characters).
  • “Tool Description” Text Area: Guidance on when the Agent should use this tool.
  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions:
    • Run as publisher (default): Agent runs with publisher’s permissions.
    • Run as current user: Agent runs with current user’s permissions.
    • Run as service account: Agent runs with a specific service account.
  • “Automation” Dropdown: Selects an existing automation with an On-Demand trigger.
  • “Input Values” Section: Lists input parameters with an ”+ Input” button to add fields. Each input has a name, required flag, and description.
  • “Output Values” Section: Lists output values with an ”+ Output” button to add fields the Agent will receive.

​
Run Agent Tool

  • “Tool Name” Field: Text box for naming the tool (no spaces or special characters).
  • “Tool Description” Text Area: Guidance on when the Agent should use this tool.
  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions (same options as Run Automation Tool).
  • “App” Dropdown: Selects the App containing the target agent.
  • “Target Agent” Dropdown: Selects the sub-agent to delegate tasks to, with an external link icon to view the agent.
  • “Worker Task Prompt” Text Area: Instructions for how this agent should delegate tasks to the worker agent.

​
MCP Tool

The MCP Tool configuration uses a 3-step wizard: Step 1: Connect to MCP Server
  • “MCP Server URL” Field: Text input for the base URL of the MCP server.
  • “Authentication” Dropdown: Selects the authentication method:
    • Bearer Token: Shows an “Access Token” field for the token.
    • Basic Auth: Shows “Username” and “Password” fields.
    • API Key: Shows an “API Key” field.
    • None: No additional fields.
  • “Connect” Button: Connects to the server and proceeds to tool selection.
Step 2: Select Tool
  • Tool List: Displays available tools discovered from the MCP server.
  • Tool Selection: Click a tool to select it and proceed to configuration.
Step 3: Configure Tool
  • “Tool Name” Field: Text box for naming the tool (no spaces or special characters).
  • “Tool Description” Text Area: Guidance on when the Agent should use this tool.
  • “Tool Start Message” Text Area: Customize what the user sees or hears when the tool runs.
  • “Run As” Selection: Radio buttons or cards to select execution permissions.
  • “Previous” Button: Returns to the previous step.
  • “Save Tool” Button: Saves the MCP tool configuration.
User Interaction:
  1. Click “Add Tool” and choose a tool type.
  2. Fill in “Tool Name” and “Description” fields.
  3. Configure tool-specific settings (e.g., App/Element selection, fields, query descriptions, authentication).
  4. Click “Save Tool” to add it.
Examples:
  • Create Record Tool: Name it “NewTicket,” describe it as “Creates a support ticket,” select the Support App, add “Issue Description” with “Ask the user what’s wrong,” and return the ticket ID.
  • Run Automation Tool: Name it “LookupOrder,” describe it as “Looks up order status,” select an On-Demand automation, define “OrderID” as a required input, and configure “ShipmentStatus” as an output.
  • Run Agent Tool: Name it “TechSupport,” describe it as “Delegates technical questions,” select the target App and “Technical Expert” agent, and provide a worker task prompt like “Investigate the technical issue and provide troubleshooting steps.”
  • MCP Tool: Connect to an external MCP server, authenticate with a Bearer Token, select a tool like “TrackShipment,” and configure the name and description.

​
4. Deploying Agents UI

Purpose: Deploy Agents into workflows and external channels like Microsoft Teams or phone calls.

​
4.1 Assigning Agents to Workflows

How to Access:
  • In record details configuration (under App settings), go to the “Buttons” section.
UI Elements:
  • “Add Button” Option: Select “Assign an Agent” from the menu.
  • Agent Selection Dropdown: Lists existing Agents or offers a “Create New Agent” link.
  • Button Configuration Form:
    • “Button Name” Field: Names the button (e.g., “Assign SupportBot”).
    • “Initial Message” Text Area: Sets the first message (e.g., “How can I assist with this ticket?”).
    • Action Toggle: Options like “Show First Message” or “Take Immediate Action.”
    • Style Options: Buttons or dropdowns for styles (e.g., Primary, Icon).
  • “Save Button” Button: Adds the button to record details.
User Interaction:
  1. Click “Add Button” and select “Assign an Agent.”
  2. Choose an Agent or click “Create New Agent” to configure one.
  3. Enter “Button Name” and “Initial Message.”
  4. Choose the action and style, then click “Save Button.”
  5. Click the button in record details to open the chat.

​
4.2 Microsoft Teams Integration

How to Access:
  • Go to the “Integrations” section in platform settings.
UI Elements:
  • “Microsoft Teams” Option: Select and click “Install Bot.”
  • Agent Dropdown: Selects the Agent for Teams.
  • Field Picker: Selects fields to save user info (e.g., email).
  • “Timeout” Field: Number input for timeout duration.
  • “Conversation End Training” Text Area: Instructions for ending the conversation.
  • “Save Integration” Button: Activates the Bot.
  • “Active Routes” Section: Shows active integrations.
User Interaction:
  1. Click “Install Bot” and follow Teams prompts.
  2. Select an Agent, configure fields, set a timeout, and add training instructions.
  3. Click “Save Integration” to enable Teams chat.

​
4.3 Phone Call Integration

How to Access:
  • In App Details, go to “Phone Numbers”; in Flow configuration, enable phone workflows.
UI Elements:
  • “Add Phone Number” Button: Opens fields for Twilio credentials and phone number.
  • “Delete” Option: Removes unused numbers.
  • Toggle for Phone Workflow: Enables phone call start with dropdowns for number and Agent.
  • Agent Configuration Fields: Includes “Training Instructions,” “First Message,” and field pickers for call data.
  • “Save Phone Integration” Button: Activates the setup.
User Interaction:
  1. Add a phone number with Twilio details.
  2. Enable the phone workflow, select a number and Agent, and configure settings.
  3. Save to allow call handling.

​
Tips for Using the UI

  • Testing: Use “Start Conversation” to test Agents before saving.
  • Descriptions: Provide clear descriptions in all text areas for accurate Agent behavior.
  • Visual Cues: Look for links like “Set up AI Service” if options are missing—they guide to prerequisite setups.
By navigating these UI elements, you can create and deploy Agents for L1 support or other workflows. Refer to the knowledge base or support for further assistance.