Whether you implement this or not, however you feel about Cursor or projects, there’s something really cool here. As people who are also building AI products, what can we learn from this? Ryan Singer, author of Shape Up, tweeted this breakdown of why Cursor works so well. Tools like Cursor hint that chat is only a small part of it. It’s really about the thing you’re working on. In coding, it’s pretty clear that you’re working on code. You want to see that in front of you. Think about other domains. If you’re working with a calendar, you want the calendar in front of you. You want to communicate with AI, but you also want to collaborate with it on the substrate in front of you. The question I keep asking myself: why not use Notion AI or MCP into Notion or Gemini or Microsoft? Why is that not the same feeling? They all have the same ingredients: models, instructions, chats, and knowledge. Maybe some of them are missing these, but they all have access to the same basic science. I came to the conclusion that at this point, it’s not about the AI. It’s about the interactions. Cursor inadvertently made it amazing for a thinking partner. If you’re asking yourself this question or you’re not convinced, try using these alternatives as a thinking partner across a lot of context. Notice what is different. Maybe it’s better for you. Why is it better? You’ll get a lot of insights on incorporating AI into products that you’re working on. ➡️ Users want to collaborate with AI on the substrate they’re actually working on.