Notion Launches Automated Meeting Agent to Handle Notes and Slack Recaps.

Notion has built an automated agent that handles one of the most neglected parts of meetings: actually sending the notes afterward.

The agent, demonstrated by Notion's Samuel Li in a LinkedIn post, pulls together relevant documents, decision logs, and previous meeting notes before a call starts. After the meeting ends, it generates a recap and posts it directly to Slack. The workflow runs entirely on Notion's recently launched Developer Platform, which gives teams the infrastructure to build custom agents and automate repetitive processes.

The announcement arrives as Notion continues expanding its agent capabilities. In recent months, the company has added Claude Opus 4.8 and Gemini 3.5 Flash to its Custom Agents feature, and rolled out an AI Academy training path to help users build their own automations.

The meeting agent operates in two phases. Before a scheduled call, it scans Notion workspaces for relevant context: project documentation, past meeting summaries, and decision logs tied to the meeting topic. This pre-call prep ensures participants have the necessary background without manually hunting through folders.

After the meeting, the agent listens to the conversation, generates a structured recap, and publishes it to a designated Slack channel. Teams can configure where notes land and what format they take, whether that's bullet points, action items, or full summaries.

The workflow is built using Notion's Developer Platform, which supports Workers (background processes that sync data or trigger actions), custom tools for agents, and integrations with external APIs. Li's demonstration walks through the full build, showing how developers can connect Notion databases, Slack webhooks, and scheduling logic to create similar automations.

Meeting notes get promised constantly and delivered sporadically. The person volunteering to take notes is often juggling participation and documentation, leading to incomplete summaries or none at all. Automating the process removes that bottleneck and ensures every meeting has a record, searchable and shareable within minutes.

For small teams juggling multiple projects, this kind of automation frees up time otherwise spent on administrative follow-up. The agent doesn't just transcribe; it synthesizes context from existing Notion pages, making recaps more useful than raw transcripts.

Notion's approach also signals a broader shift in how workspace tools handle AI. Rather than offering a single assistant that tries to do everything, the Developer Platform lets teams build specialized agents for specific workflows. One team might automate meeting notes, another might sync customer tickets from Zendesk, and a third might generate weekly reports from project databases.

The Developer Platform, announced earlier this year, gives developers declarative schemas, persistent cursors for syncing external data, and the ability to write custom tools that agents can call. Workers run on schedules or triggers, continuously updating Notion databases with data from APIs, cloud services, or internal systems.

Li's meeting agent is one example of what teams can build with these primitives. The platform supports syncing data from tools like Linear, querying live APIs, and generating assets like presentations or reports directly inside Notion pages.

Developers can start building on the Notion CLI or explore examples in Notion's public cookbook repository. The platform currently supports TypeScript and offers integrations with common services, making it accessible to teams with basic engineering resources.

Notion has not disclosed pricing details for the Developer Platform or whether the meeting agent will become a native feature available to all users. For now, it serves as a proof of concept, demonstrating how Custom Agents can handle end-to-end workflows without human intervention.