The Dash MCP server exposes a running Dash app as a set of Model Context Protocol (MCP) tools that AI agents can call directly. Because Dash callbacks are already scoped, isolatable functions with declared inputs and outputs,they can be used as tool calls so that the LLM can use deterministic, schema-declared results rather than probabilistic guesses. The agent can inspect layout, trigger callbacks, read component state, and iterate, all through natural language, grounded in real computation.

The Plotly Professional Services team are some of the best Dash developers in the world. They have adopted AI coding agents and are producing production data applications that are complex/advanced/involved Dash applications for our valued customers. They have compiled a complete set of documentation, configuration files, prompt libraries, and code samples designed to make Claude, Cursor, and Codex better and more effective with Plotly Dash. We'll walk through a live demo demonstrating how to use coding agents along with specific markdown and config files to increase the quality of code emitted and ensure it is following best practices. 

What we'll cover:

• The Dash MCP server: how it works, what it enables, and a live walkthrough of the MCP server stub.
• Several use cases for using Dash MCP
• Live demo: building a complete Dash app from scratch using Claude and the new CLAUDE.md config

Who should attend:

This session is for Dash developers who want to move faster with AI tools, data analysts, and BI engineers exploring what coding agents can do, and anyone building or evaluating Dash apps who want to understand where agentic development is headed.