Follow Along: Joe Reis Reviews Keboola MCP Server

Joe Reis, author of Fundamentals of Data Engineering, known for practical education and his YouTube content — formerly CEO/co-founder of Ternary Data — is reviewing our Keboola MCP (Model Context Protocol) Server with Claude to list Shopify his tables, run exploratory analysis, export data to BigQuery, and generate a star schema.

And you can do the same in minutes! We will go through everything here from one-click setup to best propmts to use MCP with.

What is Keboola’s MCP Server (and why it matters)

MCP (Model Context Protocol) is a standard that lets AI assistants securely call tools (APIs, DBs, services) on your behalf. Because Keboola is API-first platform it allows your AI agent to build pipeline, browse your tables or create detail documentation on the columns for you.

In Joe’s walkthrough, he connects Claude to Keboola’s MCP server, to immediately access data platform tools (e.g., get_table, get_bucket).

To see all the capabilities of Keboola MCP Server for your day-to-day data work, scroll here.

Okay, what is Joe doing…

1. Connecting Claude to Keboola MCP

Joe connects Claude and starts prompting right away. Follow his steps:

• In Claude.ai, go to Settings → Integrations*.
• Click Add integration and give the integration a name (e.g., Keboola) and paste in your Integration URL listed in your Keboola Project Settings -> MCP Server.
  
• Click Add and then Connect.

Connect to the integration - here is where you authorize your Keboola project.

• Again, in Claude.ai, go to Settings → Integrations. Click Connect. Authorize with your Keboola account.
• Close the authorization browser window and return to Claude.ai. Now you can start interacting with Keboola from the Claude chat.

*You need to be Claude organization owner or primary owner on a Claude Pro or Claude Max plan. If you use Claude Free plan, follow the Claude Desktop instructions linked bellow.

Connecting with different AI? For Cursor, Windsurf, or Make navigate to Your Profile Picture in your Keboola Project→ Project Settings → MCP Server to see all setup steps for each AI tool.

‍New to Keboola?
Set up your Keboola project for FREE.

2. Show me All Tables

Joe’s first move: “Show me all tables I can access.” Claude returns 16 tables with names, row counts, and descriptions - a quick validation that the integration works.

Try this prompt as well:

“List all tables in my Keboola project and include row counts and short descriptions.”

This uses the Storage tools exposed by MCP to browse buckets, tables, and columns (for more details, follow in documentation help.keboola.com).

3. Build me Customer Segmentation

Next, Joe asks for customer and market segmentation analysis across his Shopify data. Claude orchestrates a series of table queries and SQL steps (you can see the queries as it goes) and returns segments like High-Value, At-Risk, and more - plus useful metrics such as average customer lifetime value, number of orders per customer, and top-performing markets.

Try this prompt:

“Segment customers by RFM over the last 12 months. Return segment definitions, cohort sizes, and top 5 insights, and show the SQL used.”

4. Export My Shopify Data to BigQuery

Joe asks the assistant to run a flow that writes Shopify tables from Keboola into his own BigQuery project. After an initial hiccup with connection (yes, it is fixed 🙏), the flow completes; in BigQuery he sees customers, order_line_items, and products populated.

Try this prompt:

“Create a flow that writes Shopify customers, order lines, and products to my BigQuery dataset <PROJECT>.<DATASET>. Then run it and link me to the job.”

This leverages MCP Jobs, Flows, and Components. If you’re using BigQuery, remember you can also let Keboola host data in its own BigQuery instance for analytics performance.

5. Generate a Star Schema

Finally, Joe asks Claude to design a star schema with his data. He runs the generated SQL in BigQuery, creating facts and dimensions plus summary views in a few minutes.

Try this prompt:

“Create a star schema from my Shopify tables (customers, order lines, products) in BigQuery.”

If you want to see the full Joe Reis honest review:

What More You Can Do With Keboola MCP

• Components & Transformations: Create, edit, and run components/transformations directly from chat.
• Storage: Browse buckets, tables, and columns; surface schema and add documentation.
• SQL: Generate and run SQL
• Jobs & Flows: Launch, monitor, retry jobs, and orchestrate end-to-end flows.
• Docs search: Ask questions and pull answers from Keboola docs without leaving the chat.

New since August:

• Conditional Flows: Set up a Flow that executes steps only when conditions are met.
• Data Apps: Build Streamlit app by describing it to the chat.
• Column data types inget_table: Richer schema details exposed in responses.
• Improved SQL generation in query_data: More reliable queries and fewer edits.
• create_oauth_url tool: Get an authorization link right inside the chat for OAuth components.

Want To Try It?

Head to your Project Settings → MCP Server, connect your assistant of choice, and paste one of the prompts above. The rest is just a chat.

‍New to Keboola?
Set up your Keboola project for FREE.

‍