Notion unveils Custom Agents to automate team workflows
Notion has launched Custom Agents, automated AI “teammates” that run workflows on a schedule or in response to triggers inside its workspace.
The feature is in public beta for Business and Enterprise customers. Custom Agents run in the background and can act even when users are offline. Notion positions them as tools for recurring work that teams would otherwise repeat manually.
Custom Agents live inside Notion and draw context from information stored in the workspace. They can also work across connected services including Slack, email and calendars. Users describe a workflow once, set the conditions for when it should run, and the agent repeats the task automatically.
Common use cases
Notion outlined several initial scenarios. One is an “instant Q&A” agent that answers repeated questions using knowledge in Notion and connected tools. Another is task triage and routing, which captures incoming requests, prioritises them, and assigns them to teams. A third is automated reporting, where an agent gathers updates and delivers reports on a set cadence.
Users can also build their own agents through a chat interface with Notion AI. The system can write workflow instructions or generate an agent from scratch based on the conversation. Once created, an agent can be shared with colleagues and used across a team.
Custom Agents expand Notion's AI offering beyond one-off assistance such as drafting, summarisation and search. Notion has been adding AI features across its workspace, which combines documents, notes, projects and other collaboration tools. Other AI features, including Notion Agent, AI Meeting Notes and Enterprise Search, remain included with Business and Enterprise plans.
Pricing shift
Custom Agents are free during the public beta. After that, Notion plans a usage-based model built around “Notion credits” that customers buy as an add-on to Business and Enterprise subscriptions. Seat prices will not change.
A real-time usage dashboard will show how credits are consumed. Teams will receive alerts as they approach limits, and Custom Agents will pause automatically once limits are reached. Administrators will be able to control who can create or run agents, and can disable agents at any time.
Usage-based AI pricing has become more common among workplace software vendors as customers scrutinise AI costs and adoption. It can also shift spending from blanket per-user add-ons to a model linked to actual usage, particularly for always-on features that run without a user initiating each action.
Controls and logs
Custom Agents include logging and access controls to meet enterprise audit and governance requirements. Each run is logged, including what triggered it and what actions it took.
Permissions follow Notion page access rules. Users control what agents can access and edit, and edits are reversible because agents work within Notion's collaboration layer.
On data handling, Notion says its AI does not train on customer data and that Enterprise plans have zero data retention.
Early customers
Notion says customers including SumUp, Pigment, Pennylane and Lovable have used Custom Agents ahead of general availability. Some have used them to generate tasks from meeting notes, triage requests and produce team reports.
“Notion's custom agents have become a core part of how we operate at Pigment, with 70% of employees now using them across every team, and the majority relying on them weekly,” said Anastasia Caireac-Sailleau, Program Manager, Data and Automations, Pigment. “They handle a range of use cases including automating tasks, analysis and reporting, significantly reducing manual work for teams, while also streamlining our processes and making them more reliable and easier to scale.”
SumUp highlighted deployment at scale and a mix of workflow automation and reporting. “At SumUp, more than 3,000 employees rely on Notion to share knowledge, collaborate, and integrate AI workflows. We're using Custom Agents to generate tasks from meeting notes, triage marketing requests, and run team reports-cutting manual work and saving several hours per week,” said Frederik Severin, Head of Procurement and Artificial Intelligence, SumUp. “Because Custom Agents are always-on and have full context of our workspace and connected apps, we can explore countless use cases-from data analytics to product development-bringing greater clarity, efficiency, and collaboration across the business.”
Notion framed the launch as a response to a gap between interest in AI and practical, day-to-day automation.
“More than three years into the AI boom, there's still a gap between expectation and reality,” said Mick Hodgins, General Manager EMEA, Notion. “Our customers need AI tools that automate end-to-end workflows: connect information, automate updates, and keep work moving in the background. When agents are live where knowledge and work already coexist, the compounding effect on collaboration and productivity is incredible. That's Notion's value since day one. We can't wait to see what teams build.”