Notion 3.0 Explained | Top Features & What's Coming Soon!

Notion 3.0’s biggest practical upgrade is granular database permissions, letting workspace owners share only the rows and actions each person needs—without handing over the entire database. The example walks through a task tracker where a contractor, Sam, should see and edit only tasks assigned to them. Instead of granting full database access, the setup uses share rules tied to specific conditions (like the assignee and the entry creator). When Sam opens a shared page via a linked view, the database doesn’t even appear as a whole; it surfaces only the permitted task(s), and Sam can edit within those boundaries.

That row-level control comes with a key limitation: permissions can’t be restricted down to individual properties. If a status field is editable, Sam may still change it in a way that effectively removes the task from their view—because the filtering logic depends on the assignee/status relationship. The transcript also highlights a cleaner use case: making the assignee property view-only at the page level. In that scenario, Sam can view the shared tasks but can’t alter fields that would change what they see, making the permissions model more predictable for internal sharing.

The other major pillar of Notion 3.0 is Notion Agent, positioned as a more personalized AI layer that can follow custom instructions and take actions inside a workspace. Users can personalize Notion AI with guidance text and even “instructions” that shape tone and behavior. The agent can create new pages—such as generating a new project—then clearly indicates where it placed the content and allows undo if the result isn’t wanted. It can also apply templates, including a workflow where a user provides a transcript or images and the agent generates content (like captions), with options to store drafts in a database.

On the automation side, Notion AI formulas add a no-formula-writing path: users can add a formula property and ask AI to write, fix, or explain the formula in plain language. The example demonstrates conditional logic for task trackers—marking overdue tasks with an “overdue” tag, hiding the formula output when tasks are done, and applying styling rules like bold and green text based on status.

Notion is also expanding AI connectors so chat-based assistance can pull in context beyond Notion pages—specifically including Notion Mail—then use that combined information to compile reports. Looking ahead from Make with Notion, two features are flagged as coming soon: a team of custom agents that can operate with roles, triggers, and cycles (for example, generating weekly reports or reacting to Slack messages), and a Maps view database for storing locations and visualizing them on a map—useful for travel planning or restaurant lists.

Taken together, the updates shift Notion from a static workspace into a more controlled collaboration environment and a more action-oriented AI system—where access rules, content generation, and workflow automation can be tailored to specific people and recurring needs.