News for hackers. Stuff that matters.
Re:OpenAI reaches agreement to buy Windsurf for around $3B by Szpadel on Tuesday May 06, 2025 @09:13AM (#43904717)
For someone that never used windsurf, what features does it have that GitHub copilot does not? Reading their webpages I didn't spot any "killer feature" that would convince me to switch.

I always felt that cursor and windsurf should be just extension to vscode instead of a fork. Was there some missing functionality is vscode that was missing? Is it still missing?

There are some extensions that work in this way and allow to use multiple implementations depending on task at hand without any long term commitment.

I feel like such fragmentation is by artificial just to lock users in single ecosystem.

Re:OpenAI reaches agreement to buy Windsurf for around $3B by jstummbillig on Tuesday May 06, 2025 @09:23AM (#43904824)
It can write a lot of code, that works, better than vscode can (right now).

It's in a lot of ways the OpenAI story itself: Can they keep an edge? Or is there at least something that will keep people from just switching product?

Who knows. People have opinions, of course. OpenAIs opinion (which should reasonably count for something, them being the current AI-as-a-product leader) is worth $3B as of today.

Re:OpenAI reaches agreement to buy Windsurf for around $3B by mliker on Tuesday May 06, 2025 @11:51AM (#43906499)
Windsurf works well with Claude and Gemini models, so if OpenAI forces Windsurf users to only use OpenAI models, then it wouldn't be as useful.
Re:OpenAI reaches agreement to buy Windsurf for around $3B by throwup238 on Tuesday May 06, 2025 @04:07PM (#43909099)
I doubt they'll restrict it to their own models. The amount of business intel they'd get on the coding performance of competing models would be invaluable.
Re:OpenAI reaches agreement to buy Windsurf for around $3B by jonny_eh on Tuesday May 06, 2025 @07:59PM (#43910894)
They'll make ChatGPT the default, and defaults are powerful.
Re:OpenAI reaches agreement to buy Windsurf for around $3B by ZeroTalent on Tuesday May 06, 2025 @09:29AM (#43904901)
It's better at coding, but they are essentially paying for users.

I would also argue that the product could be built over two weekends with a small team. They offer some groundbreaking solutions, but since we know that they work and how, it's easy to replicate them. That also means they have significant talent there.

Hence, they are also buying the employees.

The code base itself is basically worth nothing, in my opinion.

Re:OpenAI reaches agreement to buy Windsurf for around $3B by koakuma-chan on Tuesday May 06, 2025 @10:51AM (#43905822)
> They offer some groundbreaking solutions

What groundbreaking solutions does Windsurf offer?

Re:OpenAI reaches agreement to buy Windsurf for around $3B by throwaway7783 on Tuesday May 06, 2025 @10:47AM (#43905764)
What groundbreaking solutions specifically?
Re:OpenAI reaches agreement to buy Windsurf for around $3B by ZeroTalent on Tuesday May 06, 2025 @10:52AM (#43905828)
AFAIK their Cascade coding flow implementation was the first done well and then copied than most.
Re:OpenAI reaches agreement to buy Windsurf for around $3B by oefrha on Tuesday May 06, 2025 @09:44AM (#43905044)
The differentiator of Cursor is it’s way smarter at basic code completion than GitHub Copilot. I pay for Cursor instead of GitHub Copilot even though I get the latter for free from open source contributions, and I made that decision after five minutes of usage after using Copilot for what, more than a year? I won’t even talk about how Cursor guesses where I’m going to edit next and makes the correct edit most of the time, just the fact that Copliot makes completions that result in unbalanced parentheses/braces all the time and Cursor doesn’t makes the switch a no-brainer; that’s not even a fucking AI problem, you just need to look around and see that function you just completed already has a closing curly brace, all it takes is some traditional AST analysis if your model is dumb. (Copilot made zero progress on that issue during my time using it, but I can’t say if that was fixed after I ditched it.)
Re:OpenAI reaches agreement to buy Windsurf for around $3B by moi2388 on Tuesday May 06, 2025 @11:54AM (#43906547)
My experience is the same. And the agent mode in copilot is terrible, it simply will stop halfway through files.

Or you chat and suddenly it wants to use the azure copilot instead because reasons.

Horrible experience.

Re:OpenAI reaches agreement to buy Windsurf for around $3B by peteforde on Tuesday May 06, 2025 @03:37PM (#43908834)
Same. Cursor might be the only tool I've purchased a year's subscription to before the end of my free trial.

I've tried just about every model on its own over the years, and yet there's something about the Cursor workflow that frequently still gives me chills when it shows me again that it had clearly anticipated what I would think next in a way I just don't experience with other tools.

Holistic seems like the right word?

If it's all smoke and mirrors as some folks imply, then it's Penn and Teller level smoke and mirrors. Beware those who tell you that they could duplicate anything of value in a weekend.

Re:OpenAI reaches agreement to buy Windsurf for around $3B by kaptainscarlet on Wednesday May 07, 2025 @03:49AM (#43913169)
Windsurf does it all the time like a wife of 40yrs completing your sentences. A good example is when you remove a function parameter. It automatically prompts to remove the arguments in all usages of the function, saving me a lot of time.
Re:OpenAI reaches agreement to buy Windsurf for around $3B by horns4lyfe on Tuesday May 06, 2025 @08:55PM (#43911173)
The feature they have over copilot is “not sucking”
Re:OpenAI reaches agreement to buy Windsurf for around $3B by johntarter on Wednesday May 07, 2025 @01:47AM (#43912603)
I'm going back and forth between Windsurf and Github Copilot right now. Windsurf's development iteration speed is much fast and features are added faster.

For example, Github only autocompletes based on what file you have opened in the current editor's tab. Windsurf indexes your entire code base and seems able to autocomplete based on what other files you have in your project. Autocomplete also spans across multiple lines and open tabs.

Windsurf's agentic tool (Cascade) can run terminal commands and read the output without opening a terminal like copilot. It can undo the agent's actions easier than Copilot. Though I think Cursor is superior in that regard, it can undo multiple checkpoints.

Still evaluating Windsurf but it, Cursor, and Claude Code are all more sophisticated than Github copilot at the moment. I'm sure copilot will catchup but by that time the other tools may have already iterated ahead.