Notion advises users not to worry as much about organizing workspaces. With AI-powered semantic search using embeddings, the system can find relevant information regardless of its folder structure. The priority shifts from manual organization to simply ensuring all data is in the system for the AI to find.
Standard APIs for human developers are often too verbose for AI agents. Notion created agent-centric APIs, like a special markdown dialect and a SQLite interface, by treating the AI as a new type of user. This involved empirical testing to understand what formats agents are naturally good at using.
The widespread use of coding agents at Notion has amplified engineering output, leading to what co-founder Simon Last calls a 'more messy and chaotic' environment. This 'productive chaos' manifests as more ambitious pull requests and non-engineering teams, like design, building their own sophisticated prototyping tools.
To fully leverage rapidly improving AI models, companies cannot just plug in new APIs. Notion's co-founder reveals they completely rebuild their AI system architecture every six months, designing it around the specific capabilities of the latest models to avoid being stuck with suboptimal implementations.
The engineering role is shifting from direct coding to 'agent management.' Notion's co-founder Simon Last no longer types code; instead, he designs end-to-end tasks, assigns them to AI agents, and verifies the final output. This represents a fundamental change in the software development workflow.
Notion's core vision has fundamentally changed because of AI. The co-founder explained their goal shifted from building the best tool for humans to *directly perform* work, to creating the best platform for humans to *manage agents* that do the work for them, using the same core primitives like pages and databases.
To personalize his email-sorting agent, Notion's co-founder didn't manually label data. Instead, he prompted the agent to ask him questions about which emails to archive. This interactive 'interview' process allowed the agent to learn his preferences and generate its own rules from the conversation.