Bring your agent to Notion with Chat SDK

Vercel’s new Notion adapter integrates with its Chat SDK, allowing AI agents to join Notion comment threads directly. The adapter maps Notion pages to channels and comment threads to threads, maintaining structured conversations automatically. It supports features like mentions, message editing, and file attachments, with default behavior triggered by @-mentions or keywords where mentions are unavailable. The integration avoids the need for a separate codebase, enabling agents to operate across multiple platforms, including Slack, Discord, and GitHub, from a single deployment.

The adapter handles limitations in Notion’s API, such as rendering cards as markdown and excluding buttons, modals, or reactions. Discussions can begin at the page or block level but not on inline text ranges. Conversation history is limited to open comments, and the system defaults to replying only when explicitly mentioned or via keyword triggers in unsupported workspaces. These constraints reflect Notion’s current API capabilities rather than the adapter’s functionality.

Vercel emphasizes compatibility with existing Chat SDK deployments, allowing teams to extend agent interactions to Notion without additional development effort. The adapter supports up to three file attachments per comment and preserves threaded replies, ensuring continuity with other platform integrations. Users can configure the bot’s behavior to suit their workspace’s Notion settings, including keyword-based triggers where @-mentions are not supported.

To implement the adapter, developers can refer to Vercel’s documentation or browse the adapter directory for setup guidance. The integration is designed to be plug-and-play, requiring minimal configuration to deploy an agent in Notion. Vercel notes that the adapter is part of its broader effort to unify AI agent interactions across diverse collaboration tools.