Introduction

This blog post provides step-by-step instructions for connecting a Watsonx Orchestrate AI agent with Watsonx Data Intelligence. It demonstrates how easily you can extend your AI assistant's capabilities using Watsonx Data Intelligence features. In this guide, you will learn how to build a Watsonx Orchestrate agent and configure it to work with the Data Intelligence MCP server.

Prerequisites

Before starting, make sure you have the following installed or configured:

• IBM IAM API Key
• Watsonx Data Intelligence Service
• Watsonx Orchestrate Service access

Following setup is done for cloud SaaS env, but it can be easily replicate for CPD as well.

Step 1 - Create Data Intelligence connection in WXO

As the first step, create a Data Intelligence connection inside Watsonx Orchestrate so it can connect with the Data Intelligence service in SaaS or CPD.

Go to Connections and add a new connections:

Give a name to the connection

You can use the key–value pair options below for different Data Intelligence instances.

For Cloud SaaS Prod:

DI_SERVICE_URL: https://api.dataplatform.cloud.ibm.com
DI_APIKEY: <your ibm iam api key>

For Cloud SaaS Dev:

DI_SERVICE_URL: https://api.dataplatform.dev.cloud.ibm.com
DI_APIKEY: <your ibm iam api key from dev env>

For CPD onprem:

DI_SERVICE_URL: https://api.dataplatform.dev.cloud.ibm.com
DI_APIKEY: <your cpd api key>
DI_ENV_MODE: cpd

Step 2 - Build and deploy Agent using DI MCP server

Go to Build → Agent Builder → Create Agent.

Under Description, enter the prompt.

You are an assistant agent.


When handling tool responses:
1. Always parse the JSON or structured output before presenting it. Never summarize without parsing. 
2. Do not output raw JSON arrays or objects unless the user explicitly requests raw output. 
3. If the response is a JSON array of objects:
 - Create a clean Markdown table with column headers taken from the object keys (id, name, catalog_id, project_id, url, etc.).
 - Include all rows from the array.
 - For URL fields, render them as shortened clickable links.
4. Keep the output concise and easy to read. Do not restate what the function did unless explicitly asked.

Select Toolset and click Add Tool.

Select “Import from MCP Server.”

Click “Add MCP Server” on the right-hand side.

Fill out the form as shown:

• Select Connection → choose the connection you created in Step 1

• Install Command → enter the command below

uvx ibm-data-intelligence-mcp-server --transport stdio --wxo 

Select Llama 405B from the model list.
Click Deploy in the top-right corner and follow the steps. You’ll see a notification once the agent is deployed successfully.

Step 3 - Chat with DI agent

Select Chat from the menu options.