The Looker MCP server: Your data's universal translator for AI

Key Concepts:

• Conversational Analytics Agents: AI-powered tools that allow users to query data using natural language.
• Looker's Semantic Layer: A trusted, governed layer that ensures data consistency, accuracy, and reliability.
• Model Context Protocol (MCP): An open standard for secure and standardized communication between applications and LLMs.
• Looker MCP Server: A key component that connects Looker's semantic layer to the open ecosystem of LLMs and data tools.
• LLMs (Large Language Models): AI models used for natural language processing and generation.
• Gemini CLI: Google's command-line interface for interacting with the Gemini LLM.
• MCP Toolbox: A set of tools used via MCP to interact with Looker models, explorers, dimensions, and measures.

1. The Problem with Traditional Dashboards and the Solution

• Traditional dashboards are insufficient in today's fast-paced environment because they don't provide real-time insights.
• This creates a bottleneck where data teams are overwhelmed with requests, and business users experience delays in accessing data.
• Looker Conversational Analytics Agents address this by enabling users to obtain answers by asking questions in natural language.

2. Leveraging Looker's Semantic Layer for Accurate Insights

• Looker's trusted semantic layer ensures that AI-driven insights are accurate, consistent, and reliable.
• The semantic layer provides a governed and well-defined structure for data, ensuring that queries are executed against a consistent understanding of the data.

3. The Future of Conversational Analytics: Flexibility and Open Ecosystem

• The ultimate goal is to make conversational analytics accessible to everyone.
• This involves providing businesses with the flexibility to choose and fine-tune their own LLMs.
• It also requires seamless connectivity to a diverse ecosystem of data sources and tools.

4. Model Context Protocol (MCP) and Looker MCP Server

• The Model Context Protocol (MCP) is an open standard that enables secure and standardized communication between applications and LLMs.
• The Looker MCP Server is the key component that integrates Looker's trusted semantic layer into this open ecosystem.
• The MCP Server is model-agnostic, allowing developers to choose any LLM.

5. Demo with Gemini CLI: Accessing Data and Creating Visualizations

• A demo using Gemini CLI illustrates how the agent can access information about models, explorers, dimensions, and measures via the MCP toolbox.
• The agent can return a SQL script, query results, and create a Look and a dashboard.
• Example: A marketing manager can analyze monthly revenue for the top three product categories in 2025 without involving the data team.
• The agent uses the Looker MCP Server to access the governed semantic layer and its defined models, measures, and dimensions.
• Looker generates optimized SQL, which the agent translates into a clear, easy-to-read result.
• The agent generates a URL to access the results directly within Looker.
• The manager can then use Looker’s charting and visualization tools to customize the visualization.

6. Interacting with Governed Data Through Different Interfaces

• A demo shows how to interact with governed data through different interfaces like Anthropic’s Claude Desktop.

7. Automated Content Creation for Developers

• Developers can automatically create content using the make_dashboard tool.
• The agent can analyze the Looker model and quickly create a sales dashboard for 2024, including monthly sales figures, sales by product, brand, and profit margins.

8. Key Quote

• "With the Looker MCP server, creating dashboards feels like magic! In just a few simple commands, you can conjure up insightful visualizations that bring your data to life, revealing key trends and metrics with astonishing speed and ease. It's like having a data analysis wizard at your fingertips!"

9. Conclusion

• The debut of Looker’s MCP Server, combined with its powerful semantic layer, is a game-changer for enterprise BI.
• The MCP Server empowers users to access data insights quickly and easily through natural language queries and automated content creation.
• The combination of the semantic layer and the MCP server ensures data accuracy, consistency, and reliability, while providing flexibility in choosing LLMs and data tools.