Notion Launches Developer Platform for AI Agents
Productivity company Notion has entered the field of AI agents with a major update. During a live product event on Wednesday, the firm famous for its team-based note-taking tool revealed a fresh developer platform. This setup boosts its own AI agents, links to outside agents, and lets groups create automated processes with multiple steps that draw data from various databases.
Notion created an orchestration system to manage AI tasks across different tools and data. The company now aims to serve as a central spot where humans and AI agents work together using multiple apps and databases, beyond simple note-taking with AI add-ons.
Custom Agents Get Major Upgrades
Back in February, Notion rolled out Custom Agents. These AI helpers manage routine jobs such as responding to common questions, gathering status reports, and handling workflows. Customers have created more than one million of these agents, according to Notion.
Before this, the agents faced restrictions. They could not link to outside data or apply custom rules. Company-built external agents also lacked direct access to Notion workspaces. Groups often used outside automation services or custom scripts on their own servers to bypass these issues.
"Historically, Notion has not focused much on developers," said Ivan Zhao, Notion's co-founder and CEO, in the live event. "That is changing now."
Notion, founded by Ivan Zhao and others, started as an all-in-one workspace blending notes, databases, tasks, and wikis. It has grown popular among teams for its flexibility in organizing information and projects.
Workers Enable Custom Code and Syncing
The new platform lets teams run their own code. Workers provide a cloud setup for custom scripts in a secure sandbox, which isolates code to avoid conflicts. This supports tasks like importing data to Notion, making tailored tools, and starting actions via webhooks. Webhooks send automatic alerts from other apps to trigger processes, all without outside servers.
Teams do not need to code manually. An AI coding agent can handle it, Notion notes.
Workers follow the same credit model as Custom Agents. Notion offers them free until August for testing.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Data syncing from external sources comes through Workers too. The tool pulls information from any API-equipped database. Examples include Salesforce, Zendesk, Postgres, and more. Data stays up to date in Notion databases.
Zhao said users can treat Notion databases as a blank base for workflows and agents.
Tools for Agents and External Connections
Workers also create agent tools with specific logic when Model Context Protocol, or MCP, falls short. MCP is a standard for AI links to external data and services.
Users can now message external AI agents directly, give them tasks, and monitor results, just like internal ones. Launch partners include Claude Code, Cursor, Codex, and Decagon. More will follow.
An External Agent API exists for linking company-specific internal agents to Notion.
Developers and agents access the platform using Notion CLI, a command-line interface. It works on Business and Enterprise plans.
Strategic Shift Toward Programmable Workflows
This developer platform marks a change for Notion. It moves from a basic app to a flexible programmable system. The setup competes with workflow automation services as companies seek to automate office tasks and internal AI setups. Combining agents, code, and real-time data in one location makes it essential infrastructure.
The update matches trends where AI firms expand from chatbots to action-taking agents across apps.
"Any data, any tool, any agent. That sums up the Notion Developer Platform," Zhao stated.