Build AI-powered collaboration on a visual canvas | ODSP904
Microsoft Developer presents a Build 2026 session on using AI to generate and organize visual collaboration boards, going from text prompts to structured diagrams and workflows. It also covers connecting AI agents to the Lucid MCP server to create and update documents programmatically and support team collaboration on a shared canvas.
Overview
The session focuses on AI-driven visual collaboration workflows on a shared canvas, including prompt-to-diagram generation, organizing boards with AI, and integrating AI agents via the Lucid MCP server to create and update documents programmatically.
What the session covers
How prompt specificity affects diagram quality
- Explains how being more specific in prompts improves the quality and usefulness of generated diagrams.
Supported diagram types (including BPMN)
- Reviews supported diagram types.
- Includes BPMN examples.
Generating cloud architecture diagrams
- Demonstrates generating cloud architecture diagrams.
- Examples include AWS, Azure, and Google Cloud.
Generating diagrams from text prompts
- Walkthrough of creating diagrams directly from text prompts.
Organizing and structuring boards with AI
- Grouping ideas by type.
- Adding instructions to guide how content is organized on the canvas.
Lucid MCP server use cases
- Examples mentioned include:
- Transcript-based diagrams
- Workshop design
- Shows how AI agents can connect to the Lucid MCP server to create and update documents programmatically.
Collaboration and review workflows
- Verifying diagram sharing with a security team.
- Enabling collaboration features for shared editing and review.
Agent-assisted diagram analysis and iteration
- An AI agent analyzes a Lucid diagram and compares changes to a code base.
- The agent produces a formatted, organized draft diagram (including swimlanes) for further editing.
Session metadata
- Event: Microsoft Build 2026
- Session code: ODSP904
- Track label in description: Agents & apps
Chapters
- 0:00 - Explanation of how prompt specificity improves AI diagram output
- 00:01:49 - Overview of supported diagram types and BPMN examples
- 00:03:25 - Generating cloud architecture diagrams such as AWS, Azure, and Google Cloud
- 00:04:12 - Generating Diagrams Using Text Prompts
- 00:06:48 - Grouping Ideas by Type and Adding Instructions
- 00:09:01 - Examples of Lucid MCP server use cases: transcript-based diagrams and workshop design
- 00:10:52 - Verifying diagram sharing with the security team and enabling collaboration features
- 00:13:21 - AI agent analyzes Lucid diagram and compares changes to code base
- 00:17:01 - Agent produces formatted and organized draft diagram with swimlanes for editing