Use Grok to Talk to Your JSON Services via CData Connect AI

Grok AI is a large language model developed by xAI for real-time reasoning, tool invocation, and agentic workflows. It enables developers to build AI agents that can reason over live data, discover tools dynamically, and take intelligent actions.

CData Connect AI provides a secure cloud-to-cloud interface for integrating hundreds of enterprise data sources with Grok AI. Using Connect AI, live JSON services is exposed through a remote MCP endpoint without replication, allowing Grok AI agents to securely query and analyze governed enterprise data in real time.

Step 1: Configure JSON in CData Connect AI

To enable Grok to query live JSON services, first create a JSON connection in CData Connect AI. This connection is exposed through the CData Remote MCP Server.

1. Log into Connect AI, click Sources, and then click Add Connection.
2. Select "JSON" from the Add Connection panel.
3. Enter the required authentication properties.

   See the Getting Started chapter in the data provider documentation to authenticate to your data source: The data provider models JSON APIs as bidirectional database tables and JSON files as read-only views (local files, files stored on popular cloud services, and FTP servers). The major authentication schemes are supported, including HTTP Basic, Digest, NTLM, OAuth, and FTP. See the Getting Started chapter in the data provider documentation for authentication guides.

   After setting the URI and providing any authentication values, set DataModel to more closely match the data representation to the structure of your data.

   The DataModel property is the controlling property over how your data is represented into tables and toggles the following basic configurations.

   • Document (default): Model a top-level, document view of your JSON data. The data provider returns nested elements as aggregates of data.
   • FlattenedDocuments: Implicitly join nested documents and their parents into a single table.
   • Relational: Return individual, related tables from hierarchical data. The tables contain a primary key and a foreign key that links to the parent document.

   See the Modeling JSON Data chapter for more information on configuring the relational representation. You will also find the sample data used in the following examples. The data includes entries for people, the cars they own, and various maintenance services performed on those cars.

   Click Create & Test.
4. Open the Permissions tab and configure user access.

Add a Personal Access Token

A Personal Access Token (PAT) authenticates MCP requests from Agno to CData Connect AI.

1. Open Settings and navigate to Access Tokens.
2. Click Create PAT.
3. Save the generated token securely.

Step 2: Install required dependencies

Remote MCP Tools allow Grok to connect to external MCP (Model Context Protocol) servers, extending its capabilities with custom tools from third parties or your own implementations. Simply specify a server URL and optional configuration xAI manages the MCP server connection and interaction on your behalf.

Open the terminal and install the required dependencies for the MCP integration using pip.

	pip install xai-sdk==1.4.0

The xai-sdk (v1.4.0) enables Remote MCP tools, and python-dotenv is used to securely load environment variables.

	pip install python-dotenv

Step 3: Generate an xAI API key

1. Create or login to xAI account
2. Open xAI API console
3. Navigate to API Keys

4. Click on create API key

After generating an API key, user need to save it somewhere safe. Recommended option is to export it as an environment variable in your terminal or save it to a .env file.

Step 4: Connect to CData Connect AI

Initialize the Grok client and configure the MCP connection to CData Connect AI. The code below establishes a secure connection and sends a natural language query to your data source.

import os

from xai_sdk import Client
from xai_sdk.chat import user
from xai_sdk.tools import mcp

client = Client(api_key="Your_xAI-API_KEY")
chat = client.chat.create(
	model="grok-4-1-fast-non-reasoning",
	tools=[
		mcp(
			server_url="https://mcp.cloud.cdata.com/mcp",
			extra_headers={"Authorization": "Basic Username:PAT"} #Base64 Encoded Username:PAT
		)
		],
	include=["verbose_streaming"],
)

chat.append(user("List the top two catalogs for me please"))

is_thinking = True
for response, chunk in chat.stream():
	# View the server-side tool calls as they are being made in real-time
	for tool_call in chunk.tool_calls:
		print(f"
Calling tool: {tool_call.function.name} with arguments: {tool_call.function.arguments}")
	if response.usage.reasoning_tokens and is_thinking:
		print(f"
Thinking... ({response.usage.reasoning_tokens} tokens)", end="", flush=True)
	if chunk.content and is_thinking:
		print("

Final Response:")
		is_thinking = False
	if chunk.content and not is_thinking:
		print(chunk.content, end="", flush=True)

print("

Usage:")
print(response.usage)
print(response.server_side_tool_usage)
print("

Server Side Tool Calls:")
print(response.tool_calls)

This code initializes the Grok AI client, connects to CData Connect AI via MCP using Basic Authentication, and streams the response in real-time. The agent automatically discovers available tools, invokes them to query your live data, and displays both the tool calls and final results.

Run the script to see Grok query your connected data source.

Query Results

The output shows Grok invoking MCP tools through CData Connect AI and returning live data from your connected source.

User can now query live data using natural language through Grok AI.

Build agentic workflows with Grok and CData Connect AI

Combining Grok AI with CData Connect AI delivers AI-powered data access without pipelines or custom integrations. Start your free trial today to see how CData can empower Grok with live, secure access to hundreds of external systems.