Setting Up AI Actions in BoldDesk: API & MCP

BoldDesk supports AI Actions through API integrations and the Model Context Protocol (MCP), enabling intelligent automation in customer support. AI Actions empower AI agents and the AI Copilot to perform tasks such as cancelling or returning orders, updating user information, retrieving real-time data, and providing contextual responses to customer queries, without manual intervention. These actions are customisable and prompt-driven, allowing them to be triggered based on natural language inputs. MCP enhances the precision of AI responses, resulting in smarter and more relevant automation.

API Actions are customisable, prompt-driven tasks that can be set up directly from the ticket details page. They allow support agents to respond intelligently to customer queries and automate repetitive operations. Through API integration, agents can connect to external systems to:

• Cancel or return orders
• Update user information
• Fetch real-time data

MCP (Model Context Protocol) Server enhances AI accuracy by providing contextual awareness. It helps the AI agent understand user intent and session details.

By leveraging this context, MCP ensures that AI-driven actions are both relevant and precise, improving the overall support experience.

Steps to Configure API Actions

1. Navigate to AI Actions

• Navigate to AI module > AI Actions.
• Here, you can view existing actions and either publish them or unpublish them by moving to draft.

2. Create AI Actions

• Click the Create Action button to add a new AI Action.
• Upon clicking, you will be presented with two options:
  • API Action
  • MCP Server
• Select the API Action option to proceed with direct API integration.
• Configure the following settings:
  • AI Usage: Specify whether the AI Action is to apply to AI Copilot, AI Agent, or both.
  • Brand Applicability: Select which brand(s) the AI Action applies to.

3. Collect Data from Users

• In the Collect Data section, define the necessary inputs or data points that need to be gathered from the user during a chat interaction.

4. Configure API Request

• In the API Request section:
  • Provide the external API endpoint details.
  • Add required headers and authentication information.
  • Ensure that API parameters align with the collected user data.

5. Test the API Response

• Use the Test Response feature to verify the API integration.
• Ensure the returned data is accurate, matches expectations, and aligns with the configured settings.

6. Configure Data Access

• In the Data Access section:
  • Specify which parts of the API response should be utilised by the AI (e.g., specific properties or the full response).
  • Apply filters to refine the data for better results.

7. Preview

Use this preview chat window to test the Action in a real-time chat environment before deploying the changes to production.

8. Publish or save as draft

• Choose Publish to go live or Save as Draft to continue editing later.

Steps to Configure MCP Server

1. Navigate to AI Actions

• Navigate to AI module > AI Actions.
• Here, you can view existing system actions and enable or disable them based on your requirements.

2. Create AI Actions

• Click the Create Action button to add a new AI Action.
• Upon clicking, you will be presented with two options:
  • API Action
  • MCP Server
• Select the MCP Server option to proceed with direct API integration.

3. Connect remote MCP server

• In the Connect remote MCP server section, input server URL to connect and manage external model actions.
• Option to add key value pair to specify the API endpoint that should be called to access data or be sent updates. You can pass AI Agent attributes in the URL or request body.

4. Load Action

• Add the action that needs to be performed.

5. Preview

Use this preview chat window to test the Action in a real-time chat environment before deploying the changes to production.

6. Publish or save as draft

• Choose Publish to go live or Save as Draft to continue editing later.

Frequently Asked Questions (FAQs)

1. What are AI Actions in BoldDesk?
   They’re automations the AI can run—like cancelling an order or updating a profile—based on what the user says in chat.

2. What’s the MCP (Model Context Protocol) server, and how is it different from API Actions?
   API Actions call your external systems to do tasks. MCP server gives the AI extra context (intent/session info) so it picks and runs the right task more accurately.

3. How do I create an API Action in BoldDesk?
   Go to AI module → AI Actions → Create Action → API Action, set who can use it (AI Agent/Copilot), choose brand(s), define the inputs to collect, configure the API call, test the response, set what data the AI can use, preview, then publish.

4. How do actions get triggered during a chat?
   They run when the user’s message matches your prompt conditions and the required inputs are provided (e.g., order ID).

5. What does “Collect Data” mean?
   It’s the info you ask the user for (like order ID or email) so you can pass it to the API correctly.

6. How do I make sure my API Action works before I go live?
   Use Test Response to check the endpoint, headers/auth, and parameters; then Preview to see how it behaves in a real chat.

7. What does “Data Access” control?
   It decides which parts of the API’s reply the AI is allowed to use (specific fields or the full response), and lets you filter them.

8. Can I limit actions to certain brands or AI experiences?
   Yes—use Brand Applicability for brands and AI Usage to target AI Copilot, AI Agent, or both.

9. What are typical things I can automate?
   Cancel/return orders, update user details, and fetch real‑time data (like shipment status or inventory).

Related Articles

• How to Create and Use Custom Prompts
• Unveiling Agentic AI: A New Era of Intelligent Automation
• Get Started with BoldDesk AI Chatbots