Notion’s rebuild for agentic AI: How GPT‑5 helped unlock autonomous workflows
By OpenAI
Key Concepts
- Notion AI
- GPT-4 API
- OpenAI APIs
- Embeddings
- Connectors and Integrations (GitHub, Confluence, Slack)
- User Activity and Retention
Notion AI: A New Era of Productivity
Notion's mission is to empower individuals to create effective tools for their professional lives. The integration of AI has significantly amplified this mission, making the process of building and utilizing these tools more intuitive and efficient. This advancement is described as providing "new electricity" to simplify the user experience.
Genesis of Notion AI
The development of Notion AI was spurred by the co-founders' early access to GPT-4. Following a company trip, Ivan and Simon dedicated themselves to prototyping a Notion version that leveraged the GPT-4 API. Within a week, they had a functional and "compelling" prototype. This breakthrough shifted the paradigm from users needing to learn complex software interfaces to simply articulating their needs in natural language, with the software emerging as a result. The development was a collaborative effort, utilizing OpenAI's APIs to transition from internal testing to general availability (GA).
Enhanced Speed and Reliability
The implementation of Notion AI has resulted in a 50% increase in speed. This enhanced velocity not only streamlines workflows but also contributes to the reliability of the platform for users.
Seamless Integration and Contextual Understanding
A significant aspect of Notion AI's development has been the focus on extensive connectors and integrations with external platforms. Working closely with OpenAI, Notion AI is now connected to:
- GitHub: For code-related queries.
- Confluence: For documentation and knowledge base access.
- Slack: For communication and team collaboration context.
This broad connectivity allows Notion AI to function beyond the confines of a user's Notion workspace. For instance, an engineering manager can query information such as "who wrote the code," "who put the design file in Figma," or "what is the PD." The retrieval of this information is made "really fast and very high quality" through the utilization of OpenAI's embeddings. Embeddings are a technical concept where text is converted into numerical vectors, allowing for efficient similarity searches and contextual understanding.
Impact on User Engagement and Retention
The adoption of Notion AI has demonstrably impacted user behavior:
- Increased Activity: Customers using Notion AI are approximately two-thirds more active on Notion compared to those who do not use the AI features.
- High Retention: A substantial 75% of Notion AI users report that they "would never go back," indicating that AI has become "integral into how they use Notion."
Customer Validation and World-Class Quality
A key aspect of the development journey is the internal use of Notion AI by the team itself, acting as their own customers. This internal validation provides significant confidence. When OpenAI itself acknowledges the quality of Notion's Q&A product, it serves as strong evidence that the company has developed a "truly world class" offering, built upon the capabilities provided by OpenAI.
Conclusion
Notion AI represents a significant leap forward in productivity tools, democratizing software creation through natural language interaction. Its rapid development, deep integrations with essential work tools, and the resulting boost in user engagement and retention underscore its transformative potential. The collaborative effort with OpenAI has been instrumental in achieving this "world class" status, making Notion AI an indispensable component of modern work.
Chat with this Video
AI-PoweredLoad the transcript when you're ready to chat so the initial page stays lighter.
Related Videos

Building long-running AI agents with ADK
Google Cloud Tech

Operationalizing AI in workflows: Lee Spacagna, Solutions Engineer, OpenAI
OpenAI

Tony Fadell: How to build real taste (and why AI makes it matter more)
Lenny's Podcast

These 3 Claude MCPs Replace Hours Of Work
Zubair Trabzada | AI Workshop

I Automated My Pre-Market Research With AI (Here's How)
SMB Capital

Stop automating tasks yourself
Dan Martell

I Gave an AI Agent the Keys to My Life (Here's What Happened) — Radek Sienkiewicz (@velvetshark-com)
AI Engineer