Future of cursor given copilot online and long context models

I think cursor has two primary advantages over competitors: the UX and RAG.

As models trend towards longer context sizes (claude 3, gemini 1.5, presumably gpt 5, the one nat friedman invested in which doesn’t use transformers) RAG will get less important. On the other end, Github has the advantage of control over the entire dev process. Some of enterprise’s PR review stuff and other things they are cooking outside of the IDE are promising, despite not have quite as good an IDE dev experience for now.

UX can be copied and I’m not sure copilot ++ will be enough.

It seems like cursor is getting squeezed. Where does the cursor team see things going and how will it adapt to these changes?

1 Like

well Cursor has one advantage is that they can add any features to the IDE without waiting for large companies to do miracles for us :stuck_out_tongue: . github is slow to release any changes because they are massive, same with copilot for example, it’s just so generic and must be so cheap and fast on their backend so they won’t burn millions of $ per hour on it, that they made it use gpt3.5 and it’s really lame :man_shrugging: with no option to even change the model it uses. Just that alone, is well worth having a separate IDE and service with customization IMHO.

Also as a client side tooling, I don’t think Cursor competes with any of the large players but complements them - so whatever will be the need of users in the client side features, they can implement.