Notion launches agents for data analysis and task automation

At the “Make with Notion” event on Thursday, the company announced the launch of its first AI agent. The agent will draw on all of a user’s notion pages and database as context, automatically generating notes and analysis for meetings, competitor evaluation reports, and feedback landing pages.

The productivity platform said that the agent can create pages and databases, or update them with new data, properties, or views. Users can also trigger Notion agents from outside platforms that are linked to the service. For instance, you can ask the Notion agent to create a bug-tracking dashboard from various sources, including Slack, email, and Google Drive.

The newly announced Agent builds on Notion AI, a pre-existing feature that could search or summarize content. But the new agent is able to tackle more complex multistep tasks, using the powers of agentic AI. The company said that the current version of the agent can perform a task that runs up to 20 minutes across hundreds of pages.

Users can set up a “profile” page for the agent to instruct it to follow directions on referencing sources, style of output, and where to update tasks and final results. You’ll also be able to ask the agent to “remember” key points as people use them. Those memories will be stored on the profile page, and users can edit them there.

In demo videos, the company gave examples of agents that could provide feedback for landing pages and update them, create a restaurant tracker, create an analysis from meeting notes, and prepare a competition analysis report.

At the moment, you have to trigger these actions manually. But Notion said that the ability to create customized agents that work on a schedule or triggers is coming soon. The company will also release a template library for agents so you can pick ready-made prompts that might suit your task.

Over the last two years, Notion has released a calendar app, a Gmail client, a meeting note-taker, and an enterprise search to get information from different sources. These are features that gave the company enough contextual building blocks to create automations. Other enterprise knowledge and productivity platforms, including Salesforce, Fireflies, Fathom, and Read AI have launched their own agents to extract and update information.