Notion's AI team prototypes features first to determine if AI can perform the task, then designs the user experience afterward, unlike traditional development where design precedes building.
The AI technology is new, and the team needs to explore what it can do before committing to a specific design or user experience.
The Q&A feature initially integrated into search performed poorly, but when used as a question-answering tool, it became a powerful feature, showcasing AI's strength in answering questions over keyword searches.
Project management varies based on the project's phase. Early exploratory phases use minimal structure with bullet points and checklists, while later stages adopt more structured task databases, milestones, and regular stand-ups.
The team matches project management tools and processes to the lifecycle of the project, avoiding over-structuring in early phases to allow for flexibility and exploration.
Shir uses AI to generate initial drafts of performance reviews by synthesizing feedback and self-assessments, saving time and allowing her to focus on providing more thoughtful feedback.
AI handles the manual work, such as aggregating data, so she can focus on the creative aspects of her job, like providing meaningful feedback.
Be precise and specific in your prompts to get the desired output. Break down complex tasks into smaller components and focus on AI's strengths in synthesis, restructuring, and searching.
Dogfooding allows the team to use their own product, providing immediate feedback and driving improvements, even if the experience is sometimes less polished than that of external users.
She initially dismissed Notion as too large, but after using it to create a job tracker, she was impressed by its functionality and decided to join the company.
Meet Shir Yehoshua, the AI Engineering Lead at Notion.
GET 21+ FREE NOTION TEMPLATES: https://matthiasfrank.de/special
Need a Notion Consultant? https://matthiasfrank.de/notion-consulting/
In this episode of my Notion Podcast, I sit down with Shir Yehoshua, the AI Engineering Lead at Notion, to talk about Notion AI and how development teams have to adjust to a new technology, where it isn't quite clear yet what it can actually do.
We also talk about best practices for prompting, what AI is actually good at (and where you should rather do the work yourself) plus her favourite AI use cases right now.
This is part of an ongoing interview series where I sit down with Notion Power Users or the people who work at Notion to look behind the scenes and show how others are using Notion to organise their work.