Showing what you mean in the age of AI

Loom’s new features—Record for agent, Loom for Mac, and MCP tools—enable teams to communicate with AI agents as clearly as with human teammates by recording detailed video prompts. These prompts capture transcripts, screenshots, and on-screen actions, eliminating the need for lengthy written instructions and reducing ambiguity in task handoffs. The approach mirrors Loom’s existing use for human collaboration, where short videos prevent miscommunication across remote teams.

Record for agent converts screen recordings into structured prompts for AI agents, including narrated instructions, screenshots of clicked elements, and navigated links. The new Loom for Mac app enhances precision with a 4K-by-default recording option, a Selection tool to highlight specific UI elements, and a Magic pen for freehand annotations. These features ensure agents receive clear, context-rich instructions for tasks that are difficult to describe in words.

Loom now supports the Model Context Protocol (MCP), allowing AI agents like Claude, Cursor, and VS Code to directly access recordings, transcripts, and metadata within existing permissions. Loom tools in Atlassian MCP integrate seamlessly, enabling agents to add timestamped comments, organize recordings, and pull context without additional server setup. Teams using Atlassian MCP v2 can link their Loom workspace to unlock these capabilities automatically.

Rovo in Loom introduces search and chat features to retrieve information from recordings without rewatching full videos. Rovo Search locates specific moments across a library, while Rovo Chat answers questions grounded in exact timestamps, drafts follow-ups, or converts walkthroughs into written steps. These tools keep context within Loom, ensuring teams can review, correct, or refine agent work efficiently while maintaining visibility into the original instructions.