# Maestro

> The Autopilot chat in Maestro provides agentic capabilities to assist you in your BPMN modeling environment for various tasks. These capabilities include:

## Maestro

The Autopilot chat in Maestro provides agentic capabilities to assist you in your BPMN modeling environment for various tasks. These capabilities include:

* Creating and editing BPMN models
* Writing expressions
* General Q&A about UiPath, RPA, BPMN, and the current model
* Suggestions for configuring activities
* Conversion from images and documents to BPMN models
* Preserving chat history and context across requests

### Connected sources

The Autopilot chat in Maestro leverages a range of built-in tools to support your modeling work. By default, all tools are enabled.

* **Web Search**—Answers questions that require updated, external information.
* **UiPath Documentation Search**—Retrieves content from UiPath official documentation.
* **Web Reader**—Extracts and summarizes content from public web pages when you provide a URL.
* **BPMN Q&A Agent**—Answers questions about BPMN standards, reviews diagrams for best practices, identifies potential issues, and helps you understand complex process flows.
* **BPMN Agent**—Creates and edits BPMN diagrams. Use this to design a new process from scratch, modify an existing one, or convert process descriptions from documents and images into a BPMN diagram on the canvas.

:::important
Autopilot for Maestro uses large language models (LLMs). Always review and validate the generated models, expressions, and explanations before you run them in production.
:::

### For BPMN modeling

Autopilot helps you build and evolve automation-ready BPMN workflows in Maestro using natural language and documents.

Use the chat to:

* Generate end-to-end BPMN diagrams from a plain-language description or from uploaded files.
* Add, remove, or reshape tasks, gateways, events, and subprocesses without manually placing every element.
* Propose and apply refactorings, such as splitting large tasks, introducing subprocesses, or improving parallelization.
* Author C# or JavaScript expressions for Script tasks, conditions, and data mappings, including complex logic.
* Explain what parts of the model do, including how data flows between activities and where errors might occur.
* Suggest validation and error-handling patterns, such as retries, timeouts, and compensation flows.
* Compare two process variants and highlight structural or behavioral differences.