Notion Adds Agent Automation, Effort Controls, and API Access.

Notion is tightening the connections between its AI agents, workspace content, and external applications. In an announcement on LinkedIn, the company detailed eight smaller updates that collectively make its agents easier to configure, embed, automate, and monitor.

The most consequential changes are aimed at operational workflows. AI Meeting Notes can now trigger an agent after a meeting note is completed, while the Custom Agents API has entered public beta. Together, these features move Notion’s agents beyond standalone chat and closer to event-driven automation.

A completed meeting note can now serve as the starting point for an automated process. Once Notion has produced the note, an agent could draft a recap, update a project page, or prepare next steps without waiting for someone to launch the task manually.

That builds on Notion’s recent work to make AI Meeting Notes find the right workspace page automatically. For small teams, the combination could reduce the administrative gap between discussing work and updating the systems used to track it. The usefulness will depend on how accurately each agent is instructed and what workspace access it receives.

Notion is also making that access more familiar. Users can share a page with a Custom Agent through the same Share interface used to add a teammate, selecting the agent and granting it access to the page as context. An agent can also be embedded directly into a page by pasting its link, which renders the agent as a chat box alongside the relevant work.

This approach could be useful for internal playbooks, onboarding hubs, support documentation, or project pages where employees repeatedly need the same assistance. It also gives workspace owners a clearer mechanism for limiting context than expecting an agent to search everything.

Notion’s model picker is now organized by provider, with favorited models shown at the top. Users can star frequently selected models rather than searching through the complete list each time.

A new effort control determines how much work Notion AI should put into a response before it begins. For Personal Agents, the control appears beside the model picker. Custom Agent owners can find it under Settings → Advanced. Notion did not disclose how effort levels affect credit consumption, response time, or the underlying model settings, so teams will need to evaluate the trade-offs in practice.

Skills are becoming easier to create as well. A user can chat with Notion Agent to turn their knowledge into a reusable skill, rather than first writing a polished instruction set. The feature complements Notion Skills, which package team workflows as reusable agent instructions. It is designed to capture prompts and procedures that might otherwise remain in one employee’s private notes or chat history.

Administrators also get a dedicated AI usage view under Settings → Notion AI → Usage. It shows how much of the workspace’s allowance has been consumed, providing more visibility into credits as agents become part of recurring processes.

The Custom Agents API, now in public beta, is the clearest expansion beyond Notion itself. Developers can bring agents into internal dashboards, customer help centers, or a Slack bot. Notion’s developer platform also supports custom agent tools that can query live data, call APIs, and generate assets, giving builders another route to connect workspace knowledge with external systems.

For smaller businesses, the update is less about any single interface improvement than the emerging operating model. Pages can supply controlled context, meetings can initiate work, skills can preserve internal processes, and APIs can put the same agents into other products. The added usage view is important because automated agents can run more frequently, and potentially less visibly, than manually initiated AI chats.

What did Notion add in this update? Notion introduced eight changes covering its model picker, AI effort controls, page sharing with agents, chat-based skill creation, meeting-triggered workflows, embedded agents, API access, and usage reporting. The updates focus on making agents more configurable and easier to integrate with existing work.

Can Notion AI Meeting Notes start an automated workflow? Yes. After an AI Meeting Note is completed, it can trigger an agent to perform a follow-up task, such as drafting a recap or updating a project page. The announcement does not specify plan requirements.

Is the Notion Custom Agents API generally available? No. Notion says the Custom Agents API is currently in public beta. Developers can use it to bring Custom Agents into external tools and workflows, but the announcement does not provide pricing details or a general availability date.