I’m going to start by saying I really like the well engineered AI handling that you have. Functionally, cursor 2.0+ is not concerning. It’s actually really good. You choices for how you presented it is what is concerning and erodes faith.
I understand you want to push an AI first workflow. I am on board with AI first workflows. I am on board with agents doing the vast majority, even the totality of the work. This is NOT me telling st the clouds about AI taking over. Nor is it about yearning for an older version I was used to. I’m all for the progress. I do not like the direction you are heading.
By putting the agent panel front and center and hiding all of the files and everything by default tells me that you fundamentally misunderstand your user base in a very concerning way. It does not matter that you also kept the option for the older workflow. It’s that you think this is what we want in the first place.
You’re designing it like this is an iPhone. Like we are regular random people who don’t want that extra noise all over the scree. You think any engineer that is actually responsible for every single line of code wants to let the AI work like it’s in a black box the whole time? That would be professional suicide. I don’t want future junior developers making that a habit, either. Whoever had this “vision" is out of touch with reality. We should be encouraging people to be aware of everything in the project, not the other way around.
I truly find this vision concerning. I think you started with a product that people already loved (vs code) and built a very good system on top of it to integrate the AI for whole-project awareness that is above and beyond anything else out there. It hit this near perfect sweet spot. And now you are overestimating the wrong contributing factors. You want to reemphasize the vs code base to highlight the AI systems. It’s the combination of a trusted, frictionless IDE with the AI agents, not one or the other.
This isn’t you understanding your user base and building a product that caters to their needs. It’s you trying to tell us what we should like. A marketing-first decision, pure and simple. Not designed for the user in mind, but designed for an inexperienced person imagines people want.
To do this while making a big deal out of calling it Cursor 2.0 tells me this isn’t some one off bad idea, either. It’s your attempt to redefine the identity of the product moving forward. You are deliberately trying to take this tool into that direction. It’s the wrong direction.
Today, I cancelled my subscription. I posted my thoughts on social media. I will be recommending against its usage on my future teams. I don’t trust you have the right vision at all. I think trying to turn coding into a black box led by AI is a fundamentally bad idea. It’s bad for literally all involved. It’s bad for the engineer. It’s bad for the company paying the engineer. It’s bad for the future of engineering. It’s bad for customers. It’s just bad.
I have personally witnessed companies implode themselves for similar hubris. Sometimes, you don’t necessarily plan for what ends up being the magic product. Sometimes, you find it. You had it, but instead of realizing that vs code itself was a major load bearing pillar, you have decided to try to overshadow it with your AI systems. And I’m doing so, you’re going to completely choke out the vs code features that brought people to you in the first place.