"What AI has done is almost given us a different gear, new electricity, to make it even simpler." This statement by Akshay Kothari, Co-founder of Notion, encapsulates the profound transformation underway at the collaborative workspace platform, as detailed in a recent OpenAI feature. Kothari, alongside Sarah Sachs, AI Lead at Notion, discussed their strategic integration of OpenAI's advanced models—including GPT-4o, GPT-4o mini, embeddings, and now GPT-5—to unlock agentic AI and autonomous workflows for millions of users. The conversation delves into the genesis of their AI initiatives, the technical challenges, and the dramatic impact on user engagement, offering critical insights for founders, VCs, and AI professionals navigating the rapidly evolving landscape of intelligent software.
Notion's journey into natively AI-powered workflows began with a serendipitous moment. Following the COVID-19 pandemic, during a company-wide retreat, co-founders Ivan Zhao and Simon Last gained early access to GPT-4. Seizing the opportunity, they sequestered themselves to rapidly prototype a version of Notion that harnessed the power of the GPT-4 API. This intense, focused effort yielded a compelling prototype within a week, demonstrating the immense potential of large language models to fundamentally alter how users interact with software. This initial foray underscored a pivotal shift: the paradigm was no longer about users adapting to software, but software adapting to the user's natural language.
Kothari highlighted this fundamental change, stating, "We've gone from people needing to learn software to just speaking in English and seeing a software emerge on the other side of that." This vision moves beyond mere automation; it speaks to a future where interfaces become inherently intuitive, responding to human intent expressed in plain language. The integration felt less like an add-on and more like a co-creation, a "combined product" born from the synergy between Notion's platform and OpenAI's API capabilities. This deep integration allows Notion AI to assist with a wide array of tasks, from drafting and summarizing documents to generating Q&A and automating complex workflows, thereby significantly reducing the cognitive load on users.
A core insight from Notion’s experience is the importance of speed and reliability in AI-driven features. Sarah Sachs noted that from internal testing to general availability, their AI features became "50% faster." This acceleration is not merely a performance metric; it directly translates into user trust and the ability to rely on the AI for critical tasks. When AI responses are swift and accurate, users are more likely to integrate these tools deeply into their daily routines, fostering a dependency that becomes difficult to reverse.
The true power of Notion's agentic AI, however, lies in its ability to transcend the boundaries of the Notion workspace itself. Sachs explained that Notion AI isn't solely confined to internal data; it connects seamlessly with external enterprise tools like GitHub, Confluence, and Slack. This expansive connectivity enables Notion AI to act as an intelligent orchestrator, pulling information from disparate sources to create comprehensive, context-aware solutions. For instance, an engineering manager can prompt Notion AI to "Create a bug tracking dashboard... populate with feedback from Slack and Notion and my emails," and the system will intelligently gather relevant data from these external platforms, organize it, and even summarize individual bug reports. This capability moves beyond simple summarization or content generation, venturing into genuine autonomous workflow creation and data synthesis across an organization's digital ecosystem.
Related Reading
- GPT-5.1: OpenAI’s Leap Towards Human-Centric AI and Enterprise Efficiency
- ChatGPT Atlas: Redefining the Web as an AI-Powered Agent
- BeeAI Framework: Extending LLMs with Tools, RAG, & AI Agents
The impact of this agentic integration on user behavior has been nothing short of remarkable. Notion observed that customers utilizing Notion AI are "almost two-thirds as active on Notion than customers that don't." Furthermore, a striking "three-quarters of them say that they would never go back. That AI is integral into how they use Notion." These figures are a compelling testament to the value proposition of well-executed AI integration. It demonstrates that AI, when implemented thoughtfully, can dramatically increase user stickiness and make a product indispensable.
The symbiotic relationship between Notion and OpenAI is also a key factor in this success. Kothari emphasized the unique advantage of both companies being "each other's customers." This close feedback loop, where OpenAI validates the quality of Notion's AI products, instills confidence that they are collectively building "something that is truly world-class." This collaborative development model, leveraging the cutting-edge capabilities of foundational models while focusing on practical, user-centric applications, sets a precedent for how future enterprise software will evolve. The journey of Notion with OpenAI underscores that the next wave of productivity tools will be defined not just by what they do, but by how intelligently and seamlessly they empower users through agentic, interconnected AI.
