I generally like what Zed is trying to become. However, all of these features and blog posts are frustraing when they struggle to keep basic editor features stable. Edit a file outside of the editor? It's not going to show up in the project pane or the git diff. Need to work inside a container because it's 2025 and we don't need to clutter our local machine with 100s of dependencies and env managers... well now all the AI stuff is broken. ACP sounds cool until you realize every single CLI in existence works better.
My wish is that Zed gets the core working correctly 100% of the time before moving on to expanding feature sets. For now I'm back in NeoVIM because it always works the first time....
My experience has been so different. Zed seems to always do the right thing for me when I concurrently edit files with other tools. Not doubting your experience or anything, but you must have a very different environment than me. Zed has been absolutely rock solid for the past year on my computer.
With the AI stuff, it feels like they invested a bit prematurely. When the Agentic editing demo came out (6 months, 10 months ago? It’s a blur), it felt right. Accepting and reviewing edits, live tracking ..etc., felt like pair programming. The ACP addition felt like a natural evolution .
With the continuous improvement in CLI tools and people’s experience with them, it feels like doing a live review or edit-by-edit approvals all feel like a drag. I personally have come to avoid using the IDE/Editor. I just kick up Claude code - plan mode, auto-accept edits. Once the session is done, switch to the editor and make necessary adjustments. I suspect people with Max subscription and “dangerously-skip-permissions” …etc won’t even care if an editor has AI integration or not.
yup, my pet peeve is there are are not way to disable line wrap. the setting that exist doesn't work and there's no way to actually disable it instead of just increasing the max characters (with set hard limit in the source code).
have a big docs or log,data file where you don't care for the rest of the line ? well too bad better have a spare editor.
this feel to me like it should should be a number #1 priority. "an editor need to nail the editing part".
The thing about "basic" here is that it's subjective. It could be that these issues only happen on your machine or that the staff (or even most people) don't need what you're asking for. Of course they should try to fix them anyway, but their backlog is enormous.
I am not getting tons of issues with Zed going out of sync all the time. I wonder if the issue is it silently having issues watching the filesystem due to open fd limits.
I've noticed that not only does it sync but it even will recognize if I rename the folder a workspace is in.
But then, I've run into a couple of strange issues that tell me there is more polish needed:
- Sometimes upon using LSP refactor, it seems like if a bunch of files get renamed, the open buffers will get screwed up somehow. Like, I'll hit save and it will write to the old filename! It's not actually a huge problem, as I can close the buffers, delete whatever excess files I accidentally create, and re-open them without error, but it is confusing as hell.
- I have indeed had issues with the file view not always updating when files are added externally, however it is not constantly. I usually just reload workspace when this happens. It is a minor frustration, but I had many minor recurring frustrations with both VS Code and Neovim before too, so I don't consider it a deal breaker.
Yeah, I have been fighting Zed to get agents to use podman on my host, but Flatpak is sandboxed and makes it almost impossible. The ideal solution would be that Zed could use podman or docker to spin up a container where agents could run free!
How does Neovim handle outside changes then? Or is there a plugin to make it work? AFAIK it doesn't reload any bufferes when files change. IntelliJ is the only other one I know that does it transparently.
In vanilla neovim you can use autoread... this does depend on a focus event like entering and leaving the pane or switching buffers. However, for my workflow which is "go to a different terminal pane and do some things then switch back" as soon as I focus the buffer it updates.
Where as with Zed it'll just keep showing the old content and in fact closing and opening file wont even change what it shows in Zed. It's really really annoying. I have to exit zed and open it again. This means if you are working with AI agents you end up having to do this often.
vim.o.autoread (default on)
When a file has been detected to have been changed outside of Vim and
it has not been changed inside of Vim, automatically read it again.
I agree with you 100%. If basic functionalities are not airtight, there’s no way I am going to deal with growing pains just because they want to get paid on AI fluffs. Contrast this with something like Ghostty.
Zed looks really cool and I would love to give it a try, but I am just too beholden to devcontainers. I know there are workarounds to use them with Zed but with many extra steps compared to VS Code and it's forks. I can't go back to not having a totally integrated container per repo.
I understand interacting with other engineers right in the editor but I dont get why so many collab tools need to be bolted on to what is basically a text editor. This will only fragment communications since its not just engineers that work in any company. Meaning you'll now have communication spread out in Slack and Zed making collaboration difficult, not easy.
I dont honestly dont get the allure of pair programming. My pair programs are the unit tests which I run as often as I can and limit discussions to Gitlab/Slack. I have worked ag FAANGs and large companies and never once pair programmed anything.
I honestly cannot think of a single software or process problem that requires real time collab in the editor. Having said that it is a cool feature and I quite like Zed as an editor.
I _really really_ want to try this feature, but only if I can selfhost the collaboration server. If there is any way to do this, it's not obvious. Given that as I understand it, lots of project details will pass through Zed's servers, I can't imagine any enterprises would knowingly allow this without some kind of SLA with Zed.
We’ve been growing and have had to scale authentication beyond what our original collab server could handle. Not many people are using collab yet, we’re prioritizing non-collaborative features (like our recent Windows release), and we’re planning on rebuilding all of this on top of DeltaDB. Fundamentally, it just fell through the cracks.
Self hosting will be a vital feature for users and enterprises though, we’re planning on revisiting it once we have a few more features settled :D
Looks great. I use some CRDTs like Loro and Automerge, are there any learnings or collaboration you're taking from them to improve DeltaDB?
Also the name, I thought it was a database with CRDTs due to the "DB" at the end of DeltaDB, but it's a version control system, I thought that was somewhat misleading, any reason why it's named so?
It's not something I am excited about, but it is something I want my IDE to do well if I must engage with it. Other remote pair programming experiences are even worse and I appreciate Zed's capability in the area even if it's not what I prefer.
A lot of my IDE choices are about extensibility and flexibility more than perfection for my preferred coding approach. After all, until I only work for myself I need to be ready to accommodate the needs of others as part of my job.
I stay away from anything Hashimoto is making after my experiences with TF & HCL, or at least the bar is much higher / projects get a lot more scrutiny
Zed is sticking to all Rust, it appears, so Ghostty wouldn't work. Also Ghostty has a lot more UI features itself like tabs etc so I don't think it's as easy to integrate in an editor which also has its own UI, so Alacritty being mainly TUI driven makes more sense.
This feels like a huge distraction from building an IDE.
A monnumentous yak shave.
I've never used or cared for multiplayer in VSCode or JetBrains. It's silly.
I've never been the pair programmer type. The only time I've needed to share an IDE is during a SEV or ridiculously complicated systems bug, and that's 1% of the time.
I like Zed. I pay for pro. I like the integrated agent stuff (though my usage model has changed a bit after 5 months of use).
I'm happy that others can type in each others' space, but this post reveals a tension here. They are building a tool for building the tool, and their own team. I think that's cool, but at a 2-3 person shop heavy polyglotted across 4 OSes and 5+ programming languages, this is not what I really need.
What I'm looking for is a snappy tool (check) that lets me explore, understand, modify code at a next level (marginal). And I want it to not only be snappy by virtue of execution efficiency, but cognitive load. I want the less-is-more experience. I don't need it to do Swift, Kotlin, or Python, because there are bespoke IDEs for each of those that focus on the environments where I deploy them best. What I mostly want from Zed is the ability to see the outline panel at the same time as the directory panel, and to separate the search outline from the file structure outline. I spent too much time toggling views in Zed.
And then toggle between that and the AI agent? Zed is a workbench that lets you put two, and only two, tools on your "workbench" to either side of your text workspace. I want to put more tools on my benchtop.
Maybe I'm old, jaded, stubborn and paranoid, but something about a coding editor that is controlled by a company is off-putting to me. It's even more off-putting when you add Zoom, Slack and everything else into said editor.
…you’re free to use other editors? People like Zed. They like IntelliJ. They like VSCode. If you have an aesthetic preference against all professionally maintained IDEs, I think you’re in the minority.
There is a lot of complaints about Zed in the comments here. I don't think that they are "hate", per se; they all definitely care about Zed and want it to succeed.
I daily drive Zed for work across several languages and I love it. I use a lot of its features, like the git interface, agentic editing, etc. I might even consider paying for Pro in the future if I want unlimited edit predictions.
However, all of these complaints are fully justified. I think Zed is a massive undertaking, only one that a VC-backed company has the capital to do. iirc, it requires 70k lines of Rust just for the cloud part [1]. I cannot fathom the amount of fundamental infrastructure they have to get the editor functional at all. That doesn't excuse all of the papercuts in Zed though.
If I were Zed I would do the following:
1. stop all work on future features, like DeltaDB etc. They all seem extremely cool but they won't meaningfully contribute to increasing Zed adoption or fixing its issues.
2. remove all agentic editing features. if Zed tries to simultaneously become the world's best agentic editor and a good general-purpose text editor, it will fail at both. Keep around ACP so users can still use other agents, but remove all of Zed's built in agent stuff.
3. fix literally every papercut. Triage every single issue and go through every PR, even if it will take half a year to do so. People won't switch to Zed until it's perfect, and the existence of this many issues means it's not perfect enough.
4. make extensions actually good. Every programming language, library, etc. has it's own ecosystem, and many such ecosystems mainly rely on VSCode extensions for advanced features. Zed needs to be extremely extensible like VSCode is; obviously its architecture makes this slightly harder, as it's nontrivial, for example, for extensions to render their own GUI, but there are a lot of low(er)-hanging fruit for extensions that need to get solved. People will only switch to Zed if they can get a similar breadth of ecosystems.
Of course, this won't happen, and given that none of these will really make them money, Zed has no incentive to focus on these, especially given the amount of time they would need to do this. But I think that if Zed can't nail the core experience, it won't get anywhere.
> Despite attempts to make Atom—an Electron application—more responsive, it never reached the performance standards the team yearned for.
This feels like an attempt at deflecting blame. VSCode is another Electron application that ended up having better performance than Atom. There's another Electron adjacent application that has good performance, the one you're probably using right now to read this page.
What are you saying here? It is true that VS Code is less bad in terms of responsiveness in comparison to Atom. Zed, however, is written in Rust (i.e., not Electron), and I would guess it is at least an order of magnitude more responsive than VS Code across every possible scenario.
Web technologies are an unrivaled technological marvel for what they are, but it is disingenuous to imply they represent anything near the peak of what we are capable of in the context of performance.
It depends. It definitely opens faster and the general UI seems a bit faster, but open a largish file (a few MB) and VSCode will easily out-perform Zed because it doesn't have that fancy CDRT thing.
In my experience VSCode is plenty fast. Use it with no extensions and you will have zero problems with performance (though memory use isn't great). The real problems come when you have extensions, especially because it's often impossible to attribute performance issues to them because they can often do a lot of work all in the same "extension host" process.
Based on this comment [0], they're building DeltaDB as a version control system which uses a CRDT, so I assume even in single player mode, the file will instantiate its own CRDT for fine grained tracking of changes.
Zed is lovely and I hope it becomes super successful but this kind of mass collaboration might be ok for meeting minutes... maybe. But thinking of it for coding it gives me shingles. Code by mass live committee. Yikes.
I think it's a fun and interesting idea for training junior engineers and possibly for other use cases. Suggesting alternatives to (perceived) bad practices the instant you see them could be helpful for many people, and also save a lot of future time for reviewers.
I could also see it as a potential productivity aid. Person 1 sees Person 2 is writing something and they don't want to be seen as idle, so they start working as well. This might sound oppressive but a lot of people who struggle with ADHD/procrastination/akrasia actually receive great benefit from that structure. Similar to that startup that forces you to code while screensharing with a stranger in order to push you to work, or people who code in cafes/libraries to be more productive.
As long as it's not an organization requiring it for senior engineers, I could see promise to it as an eventual common new paradigm.
You can use remote screen control with Zoom to train up juniors. More often than not, you'll want to share the whole screen so you can show them webpages and other things that do not live in an IDE
Pair programming can be really great. Or horrible. Depends entirely on the people.
This would be good for code-walks too though. Instead of having to share your screen and hope the video comes through well. Everyone can follow along in the comfort of their own editor.
it's probably subjective, but I find these collaboration features can be overused for this kind of thing.
If someone is walking me through something, I just want to see what they see so I can focus entirely on what they're saying and no part of me is distracted by having to follow along or seeing other code.
I know typically these collab modes have an auto follow feature, but it's not as simple as just read only video being streamed to you, there's loads more ways it can go wrong and add noise / distraction that provides no benefit.
Let's lean into the chaos and see what it might give us. Imagine a production application deployed directly from a non-version-controlled directory. Anyone on the team can edit the files, at any time. Insane? Probably. The disadvantages are easy to see.
But the positives are really compelling: 1. make small, granular testable changes; 2. use feature toggles; 3. refactor intensely and concurrently; 4. always work on the latest code; 5. use in-code documentation instead of GitHub/etc workflows; 6. explore continuous, incremental, hot-swappable code deployment.
Doesn't thought of ditching all the wasted motion and ceremony around logging async work and just coding sound glorious? I'm actually not a "move fast and break things person" usually. But the idea of moving so fast that broken things will only stay broken for a tiny fraction of the time is pretty compelling. There is also an intensity that comes from real-time interactions where a team needs to reach consensus quickly.
Pair programming usually has a single "driver" on the keyboard to keep things controllable. Here, everybody is driving: "dozens of cursors are concurrently editing the same file in real-time."
a few years ago our company used Screenhero which allowed editing with multiple cursors while screensharing.
The experience was actually quite nice for two-three people but we always had the "ok let me type now" flow. Multiple changes happening at once sounds hyper distracting.
my understanding is that this is the dynamic in modern college classrooms. everyone opens a big shared google doc for notes and they all collaboratively edit a set of notes in real-time.
or at least that's what i've heard, no idea if they actually do it.
it is nice to see a crdt backed editor tool for markdown and code though. gdocs markdown support has been lacking for years.
> or at least that's what i've heard, no idea if they actually do it.
Yeah, we used to do that back when I was in college. It's only for certain classes and most people usually kept their own notes too (or instead, why write twice). Or some classes ban laptops so you'd write on paper anyway.
Yeah, people are different, but a lot of this difference results from various constraints, such as cultural practices around collaboration or technological options. Many of these limitations probably shouldn't be locked in by our tools or norms.
When learning a new way of thinking or moving (i.e. martial arts) people often really benefit from high-bandwidth, low-latency, shared-viewport-onto-reality interactions. Watching someone's cursor move while they talk is one way to get a window into their problem-solving toolkit.
These are definitely some interesting features, though not sure I'm in any position to take advantage of them at all.
The multi-user editing is kind of cool... there's an ANSI art tool (PabloDraw) that you can run a host session so multiple artists can create text art, and I thought back when I first saw it, that it might be cool to be able for multiple editors to work on a project. I've used some of the collab stuff with VS Code, but haven't done enough to even begin to compare.
Not to mention that in a lot of workplaces, self-hosting or otherwise layers of bureaucracy stand in the way.
I would love to see collab servers take the same path as LSPs in being standarized and integrated across various editors and IDEs. I would love to work more closely with my VSCode peers, for example. Of course some features may be outside the standard and only supported with likewise editors, e.g. voice chat perhaps, but having shared cursors and a text chat would be a good start.
This is what I'd like to see as well. These collaboration tools are really good, but I barely use them because they always assume that you and your team are using the same editor. Most of the time that's just not the case, so I've used them a handful of times but beyond that there's little opportunity.
It's probably not an issue the Zed team will experience as they're all naturally using their own editor. Hopefully it's on their radar though.
> because they always assume that you and your team are using the same editor.
Network effects are probably a strength for a company, not a drawback (which it is for the user of course). Even VSCode has some notion of network effects, such as their proprietary extension store.
I use Zed instead of my normal text editor because it opens instantaneously. But I don't write code in it, and still use my IntelliJ + IdeaVim with Claude Code and Codex in a separate terminal in Ghostty.
But Zed is an insanely fast text editor to open text files in. Just double click and it's on the screen. Love that. Maybe over time I'll do more in it.
I do use (and am in the Backers.md of) neovim. But the GUI versions open much more slowly than Zed. Zed is near instantaneous. I do often edit text in neovim by opening a terminal window and doing things, but the experience of browsing a large log file and filtering through it is much nicer in a GUI text editor, so if I ever double-click or open a JSON file I want that to happen instantly.
I can't stand Nano. I find it impossible to use because my brain is strongly vim-bound.
If you've been a developer long enough, you might recall the teletype package for Atom—both built by Zed's founders.
I first experienced this in SubEthaEdit in 2013 or so, but it has been around since the early 2000s:
Appropriately working together on a truly collaborative tool, Martin Ott, Martin Pittenauer, Dominik Wagner, and Ulrich Bauer of Technische Universitat Munchen won the Best Mac OS X Student Project for Hydra 1.0.1, a Rendezvous-based text editor that enables multiple people to contribute to a shared document. (Adam and about ten other attendees at MacHack used Hydra to take notes during this year’s Hack Contest.)
It seems like the "unlock" here that makes it different this time is organization-wide sharing.
People have been doing collaborative text editing since the 60s actually! See, The Mother Of All Demos[0], referenced in our first blog post[1] :D
I'd say CRDTs are also a big change. CRDTs make live collaboration much more robust for all parties involved, and they only started to reach maturity in the mid-late 2010s
SubEthaEdit was a very inspiring software project for me. The fact that a small team could, in a few months, produce an amazing app that solved real problems and gained notoriety was amazing.
As time goes on it feels like much of the low hanging fruit opportunities in software is disappearing faster and faster. I'm also a fan of Zed and everything they're doing, but it's notable that shipping next-gen editor software takes a lot more developer effort now than it did in the 2000s.
> it feels like much of the low hanging fruit opportunities in software is disappearing faster and faster.
Yes I agree but so many things that might seem "done" (and in someways I think software/SaaS as an ecosystem is "done" compared to where we came from).
BUT - so many companies just bloat themselves and their products. I think the end of ZIRP is going to have an effect on that (more enshitification / rent seeking for sure) and I think there will be an opportunity to iterate and make copyware that doesn't take the higher development efforts.
We really need a winning electron alternative that is more resource friendly. That, IMO, will be a big game changer and I know there are lots of promising alternatives already.
I'd forgotten all about it but SubEthaEdit was such an amazing tech when we were using to collaborate internationally back in about '04. It went off my radar but I am glad to see its still available as a free app.
I tried the collaborative features to pair program with a colleague a few months ago, but it was bad. It was very flaky in establishing a connection. In the cases we were able to establish a connection, the voice chat would not work. We tried to make it work for a couple of days, and then we gave up. Has there been lots of work in the past few months on the collaborative features?
It would have been good to include a link to the collaboration docs https://zed.dev/docs/collaboration in the article. There were a lot of links in that article and a lot of assumptions that I knew how things worked. And I daily drive and like Zed, but I had so many questions.
I was very surprised to find a "forum" integrated in Zed when I first opened it. But to be honest, it is not something I ever felt the need for and overall I don't like having this in my text editor. So far it never got in my way and that's a good thing, I hope it stays that way :)
I’ve been using “The Notioning” for the last few years to refer to the convergence of tools like slack adding notion like features, clickup adding notion and slack type features, and so on. There seems to be a stable set of features that retains teams in an org
Looks a lot like Google Wave. This is interesting for some things but I don't think coding is it, for the same reason that IntelliJ's CodeWithMe doesn't work for pair programming. And apparently one of the guys is a former Pivot, so it's a little surprising.
Pair programming is Two People One Cursor. A critical aspect of it is you're both looking at the same lines of code and working on the same problem and following each other's thought processes.
CodeWithMe (and it seems Zed) is Same Codebase, Same Day. There's no shared focus. You edit stuff, I edit stuff, maybe there's overlap. But this isn't much different from doing separate git commits.
So far the only remote pairing tool I've found that works competently is pop.com.
With zed you can also share your screen in the editor which makes it a bit better, but still you can't take control of the other machine.
IMO if you only care about coding doing it in the editor is the best approach, you get zero latency and have all the context that you need (most of the times). But if you want to do more, like opening the browser for whatever reason, or teaching how to use a specific cli, etc, then taking control works better.
If you liked pop you might like gethopp.app, which is an OSS pair programming app (full disclosure I am the co-maintainer). Unfortunately because we have chosen tauri for the frontend we can only support macos and windows, but I am working on a solution for Linux too.
Pair programming so often degrades into one pilot and one person just sitting there trying to catch mistakes. When those mistakes are caught they are mentioned taking the first persons attention away and breaking flow.
In contrast how I like to work, with similar level people, is to work at the same feature in the same codebase at the same time. We either sit next to eachother, or have a remote call, where we continuously talk through what we want to achieve. Sometimes this results in one person writing ahead (code, docs, doesn’t matter) and the second sweeping behind it and cleaning it up. Two cursors, one source. IntelliJ even manages to keep authors correctly in git.
The other mechanism is where we work on different parts of the code base at the same time. Either main code and tests, or split across interfaces and implementations. Because this happens on the same machine the iterations are way faster as they are local and incremental.
This basically saves the whole dance of creating branches, pulling/pushing, the fixing typos etc.
I'd say it's medium gray on white, which is not too bad in my subjective opinion. I have seen far worse. Light gray on white was "trendy" for a while and dark gray or dark green on black has always been popular among the edgy crowd.
i'm now guessing the software engineering universe is becoming like the lawyer ones
1. big corporate shops / vc funded ones - many tens of programmers working on features (this is where zed collab features might be needed)
2. bespoke high productive small teams - less than 5 product programmers in a company e.g basecamp - these would be your bespoke law firms
3. the indies (injury lawyers) -> 1 - 3 programmers chunning out products at scale or eating of one product + maybe with help of A.I
for 2 & 3 - a lot of stuff being shilled is not needed. a legal pad + some notes that can be posted via a google doc is all that's needed. Jira isn't needed too
To me it feels like 2 different set of products, what they are showcasing here seems very similar to slack/teams.
Of course, it would be awesome to have a faster and open source slack, and if I can take notes on the same style as my editor great. So I guess, it would be nice to be able to embed zed in another product.
I think this would be appealing for a company that it's core product is code, like zed, but I do wonder if other companies even need this functionality.
Technically really impressive. In practice, completely unpractical in any medium to large organization. And although I adore Zed's speed and reliability, I still don't understand why we need these features at all.
Because "maximum editor" was achieved with Visual Studio Code (which is why all these new editors look and feel like VSC), the same way "maximum toothbrush" was achieved with the electric toothbrush. But the toothbrush industry had to keep going, so now we have a $400 bluetooth connected toothbrush-as-a-service that monitors your brushing habits to optimize teeth cleanliness, with brush heads that cost more than 5 regular toothbrushes.
Zed and the current crop of AI editors (including VSC itself) are that toothbrush.
Here is where I've settled on for Zed. I initially thought it might be a Sublime replacement for one-off files, but it seems it's geared towards projects. It's not as powerful as Jetbrains (RustRover, PyCharm etc), but is much faster. So here's how I'm using Zed:
- On my Tablet, which is too slow for Jetbrains IDEs to run smoothly
- On certain projects I have which choke Jetbrains IDEs. (Due to macro use maybe?)
I think its' a much nicer experience than VsCode, which I admittedly haven't figured out to run in a project-oriented way.
I'm also trying their GPUI library, but am in the early stages, so can't really comment on how it compares to EGUI.
Zed is the only modern IDE-like editor which is fast enough to replace (n)vim for me. I plan to use it for more and more projects, but I've had minor issues with it's Vi-mode.
I'll always remain someone plugged into vim because I need it sometimes when shelled over a terminal. Editing files over SSH can work with editor support, but is often less reliable or fast than jumping through whatever hoops I need to to get an SSH connection once and then doing everything from there.
I just have a few plugins which help. Mainly the LSP for gotodef and popovers for type info, etc. This was what finally made me transition to neovim. Also a tree viewer, Startify, and :Rg for ripgrep integration. Those are my big ones.
Sadly my workflow of using `!` to get back to my terminal and things like `!make` or `!cargo build` is fucked in neovim. So I do a lot of ctrl-z and the a lot of killing stopped processes I forgot I suspended. I've complained about this in various threads and chats, but the developers aren't interested in letting us use the old vim `!` which is super lame.
> Sadly my workflow of using `!` to get back to my terminal and things like `!make` or `!cargo build` is fucked in neovim. So I do a lot of ctrl-z and the a lot of killing stopped processes I forgot I suspended. I've complained about this in various threads and chats, but the developers aren't interested in letting us use the old vim `!` which is super lame.
I'm sorry... what?? I still use vim as I haven't found a reason to jump to neovim, but you're telling me external commands don't work properly? That's wild.
They "work"... but instead of switching back from the alt screen and giving you your terminal back, they open a gimped little window within vim which makes it hard to navigate, copy, and search for things in the output.
Surface tablet, so it's like a laptop. I can kick the keyboard off when I'm reading stuff, to allow room for food and drinks. And I do drawing/notes with the pen.
Another way of stating this: It's a general purpose portable computer; not specialized coding PC.
Very cool idea, and helps promote owning your own data, and it being highly interoperable (plain text!)
I do wonder if we need a term for shoe-horned dogfooding though. Like sure, you can do this. You could do this in Figma! Or in Notion! Or in LEETCODE if you wanted to.
At least with Zed though, its plain text. If you find another way to collab realtime on plain text, you're not bound to 1 vendor.
Your company has a user pool, you sign a BAA or start working with a partner company that has their user pool. Instead of creating slack accounts in both you can share external slack rooms that only people that are invited in/from their respective orgs can join without having to co-mingle employee user pools.
But why would external partners want to look at your code? I guess if you're also integrating with them? But generally you just give them repo access instead. For Slack, it's different as messaging is a core feature to collaborate between different people in different companies, but looking at code is a very specific use case.
I could imagine that in ten years git will feel strangely slow and ceremonial. Why not just continuously work and continuously deploy live-edited software
I feel the opposite way, that git branching and merging will become a bigger part of the job as more code is written by agents in parallel and then accepted by other agents or humans.
for now yes absolutely. but I’m already hearing rumblings that some people are having luck letting multiple agents edit the same directory simultaneously instead of putting changes through PR merge hell. It just needs coordinations tools, see https://github.com/Dicklesworthstone/mcp_agent_mail as one (possibly insane) prototype
for example it’s not out of the question that we could end up with tooling that does truly continuous testing and integration, automatically finding known-good deployments among a continuously edited multiplayer codebase
we’d have to spend a lot more energy on specifications and acceptance testing, rather than review, but I think that’s inevitable - code review can’t keep up with how fast code gets written now
Having tried a janky version of this myself with a NOTES directory, I am very bearish on this being a better workflow than just improving the ui wrapper around git worktrees and the isolation that provides.
Codex already has a fantastic review mode, and gemini / claude are building tools around pr review that work no matter how that pr was produced, so I think this interface is going to get baked in to how agents work in the near term.
Yes, but does it need all the ceremony surrounding it? If, every time I saved the file, the changes were analyzed and committed to git, and a useful commit message included, and commits squashed automatically and pushed and tested and tagged (using magic, let's say); if the system existed in the background, seamlessly, how would our interactions with source control and with other developers look?
automated commit message will tell you the "what" not the "why".
In any circle of "what makes a good commit message and why even do it" discussions, invariably the recommendation is to explain the "why" and leave out the self-evident "what".
If your stance is that commit and commit messages can be automated away then we might as well not even have them.
I don't share this view, but yeah in this world we don't need AI to do things that shouldn't be done in the first place.
increasingly, the automated systems have access to the original ticket or bug report, and maybe even the conversation while implementation is happening. They can record the “why”
I love it when I have a tool that’s “done” but the software I work on in my career is never, ever done. It’s almost like there’s two different things we call “software”. there are tools like, idk, “curl” where you can use and old version and be happy. and there are interactive organizations in the world, like, eg, Hacker News, which mutates as the community’s needs change
Software for evolving business-needs is the same for me. What's insightful is that we (I) take continuously evolving software as just that: evolving. It's a defacto virtue to continuously tinker.
Doing away with check-ins entirely is the extreme end-game of that pov. I'm in product and every day and every week yes we very much continually change the product!
But I'm growing less convinced that the natural end-state of this methodology produces obviously better results.
it doesn't work quite well for complex projects that require integration with other teams/software.
You would need to either have separate versions running at the same time or never do breaking changes or devise some other approach that makes it possible.
I think that’s a tooling problem. Maybe we do end up running a lot more versions of things in the future. If we believe that code has gotten cheaper, it should be easier to do so.
Imagine if someone clicks the deploy button when you're in the middle of typing something and then the service goes down due to a syntax error. To prevent this, we will need some sort of way to set a global lock to indicate that "I'm not done typing yet" and you can only deploy once everyone has released this lock.
Or you don't deploy unless it makes it through at least testing, and a build started while someone was editing the code would probably fail fast unless you coincidentally hit the button right when it's valid, but wrong, code.
I have been trying to figure out how this works in concert with Git (or SCM in general). Is one of the developers in the session merely responsible for it?
I mean, you have the same "problem" when peer coding in person. Whoever is officially working on the fix will commit it. I've helped devs get around a hump for ages, you don't get "credit" for all the work you do. It's why I hate most ticketing systems (when management starts to ask why your tasks fell behind), they don't let you correctly track multiple people when they work together.
The collaboration tools built into Zed have basically existed since the product launched. It is one of the primary drivers behind the product - they wanted to build the best editor for remote code collaboration.
I think the most difficult question are going to be how do you constraint that core feature without ever wanting to add more to it?
For instance, collaboration is a huge topic. You can have coding collaboration on the file, and that would be basic and appropriate, you can then replicate slack and you'll have chat rooms, which is entering creep territory, but it's natural! Then soon the chat room will need to link with issues and you can now have TODOs linked to some kanban board and we should be able to speak while we code on the same file! And this goes on and on.
It's exceedingly rare that the organization found hard courage to specifically avoid features that looks like easy pickings for the purpose of avoiding them.
Ever since Apple Silicon macs came out, it's been a real struggle. Almost 6 years later and there is still nothing on the market that:
* Has the same level of performance
* With the same or better battery life
* With the same quality of screen
* With the same quality of speakers and touchpad
* Runs as quiet or as cool
as the Apple Silicon macbooks. If you add in "needs to be able to run Linux" your choices go down from maybe 1 or 2 to 0.
They all have some sort of compromise. Either the speakers, screen, keyboard, touchpad, build quality, battery life, or thermals.
I have a Surface Laptop 7 with the Snapdragon X Elite, and it's pretty close. Checks the boxes for Screen, build quality, and touchpad. Loses out on speakers and battery life, and the fans need to run a lot more than my M4 Pro MBP does. It also loses on performance, and it doesn't run Linux. Windows on Arm also still has a lot of little quirks and bugs that start to become daily annoyances.
It's incredibly frustrating. I want, essentially, my 14" M4 MacBook Pro, but in a Linux laptop, and there's no OEM out there that's fulfilling that need without compromises.
Apple keeps pulling ahead in silicon and every other laptop OEM is just being left in the dust, shrugging their shoulders, and putting out the same old 1200p 16:9 plastic garbage they have always been putting out.
As long as I don't have to use, feel free to include it. It is really not essential feature for editor.
I run update and Collab requires you to sign in... which again, it is fine if you want it. I don't, so it can be dormant, icon is really tiny, doesn't take much space.
The feature of Zed that is most annoying yet essential is frequent updates. Pretty much daily when I switch to Zed window, I can expect update and restart, which messes up my window layout, so this is annoyance. Getting updates and knowing you guys are shipping good stuff is what is essential.
I think integrating terminal ai's is great move and useful. Sometimes I use it like that, often I use it in terminal (like the outside of the editor terminal) and switch to editor to review or update stuff. Same with git. I am old-fashioned.
Looks very cool, and of course it's nice to basically have Slack inside of Zed.
But personally, what I want in a Code Editor / IDE, is to be the very best experience at writing code and working with code projects. That is what will save me time and make the coding experience better.
Collaborative features are nice but not essential since there are other tools out there. It's not likely to move a team away from Slack (though if it's self-hosted, it stands a chance).
I'm not yet at the point where I can rely solely on Zed for python coding. I mostly use Zed because I like new initiatives, especially open source ones, and it's fast and responsive. But PyCharm is still better for python development at this point, with its one black mark being endless indexing on large codebases / dependencies, and I find myself falling back to it regularly. I would argue that the priority should be to achieve parity as a _code editor / IDE_, and then we can talk about other shiny new features.
I love Zed. I, mostly, love the direction they are taking the editor in.
But. There are now two times I see Zed going in the wrong direction. The AI integration was one. This feels like the wrong direction again.
I never really liked the AI integration. It felt off to me. I do love coding with Claude and I think I know why. It presents the "information I need to know" in a way my puny brain can handle it. Colored diffs. Summaries of what happened. It isn't perfect, but it has been incredibly productive for me. I never got that from Zed's AI integration; perhaps this has been improved, but I was up and running with Claude in a way that I never was with Zed.
This write-up sounds like "slack in my editor." If it is that, I hate it. Slack has destroyed company culture and communication. People, who are inherently lazy (I'm an old Perl programmer, so I can say that), have stopped thinking carefully and writing carefully, and in that void just throw the first thing in their head into a slack channel and think that is "collaboration" and "communication." It's toxic.
For example, this comment rubs me the wrong way: "Staff members hop in, volunteer to show off a cool feature or bug fix they worked on, and get real-time feedback from the rest of the team." I don't think our human brains work well with "real time feedback" UNLESS we have the information presented in a way that gives us massive clues on what's right and what's wrong. Reading a wall of text is not the way. A colorized git diff, or a video, or an entirely new way of presenting information might make real time feedback possible, but I am highly skeptical a text editor is the way or place to do that. And, I'm an emacs user and love text UIs, don't get me wrong.
Do I want to have "generalized one off rooms for things that don't fit anywhere?" I definitely don't want that. I want you AS THE AUTHOR to be really intentional about what's important and fit that into the proper channels. I need to know that information, but I don't want to know about, nor have the unspoken expectation that I SHOULD have known, about the other stuff. And, I want "managers" (if that still exists) to be carefully thinking about those channels and how the company is organized and push that structure down to people in the organization.
As Zed is the office, having one off rooms instead of in person coffee time feels very dangerous. That's the world a lot of people live in, but I don't like that office.
If this comment is the guiding light, then I'm worried: "We're building toward a future where collaboration is continuous conversation, not discrete commits—where every discussion, edit, and insight remains linked to the code as it evolves, accessible to both teammates and AI agents." I'm human, I have kids, I have other interests. A continuous conversation is impossible for me. I want discrete ideas, and right now, discrete commits and PRs are better, IMHO, than what I hear here. It's hard, but setting the expectation that to be successful I need to be paying attention to a river of information flowing by seems like a bad idea to me. I don't buy that Zed solves the problem of hiding the pieces of information that I don't need to see.
Oh hey! I have an idea. Why not use AI to summarize those conversations into discrete pieces! </joke>
I do love Zed. It is the best GUI editor out there. I know they will get it right. I just am skeptical about this direction and feel it misses the forest for the trees.
Man, Im like the total opposite in terms of preferring the Zed UI vs claude code. I really try to avoid raw claude when possible. I very rarely pull it up to do concurrent sessions when I have Zed open already working on something else. Or if I need to do something quick while in the CLI in a random directory. Otherwise, I think just the "files modified" feature is worth using Zed as the primary interface.
You make some good points, and I need to revisit Zed+AI to see where things are at. This probably proves you are a better developer than me.
But, also, after reading your comments, I'm just not sure I need an "editor" anymore. I love that I can npm install claude anywhere. Zed does not exist for ARM servers yet, but I can install claude there, and it can troubleshoot my database connections, and edit code, and grep files. Those are all the things I used an editor for, because an editor has better ergonomics than using the CLI. I'm sad to say "misspelled prompts" might have better ergonomics for me.
If I were still hardcore avoiding a GUI Id probably be in the same camp. But I moved to VSCode as my primary interface a long time ago, and Zed is just a better version of my workflow when comparing against that.
100%. I also moved to VSCode years ago. Then, I got disillusioned by the performance, and by M$ telemetry. Zed is so much better in so many ways if you have to go with a GUI for all those reasons.
I generally like what Zed is trying to become. However, all of these features and blog posts are frustraing when they struggle to keep basic editor features stable. Edit a file outside of the editor? It's not going to show up in the project pane or the git diff. Need to work inside a container because it's 2025 and we don't need to clutter our local machine with 100s of dependencies and env managers... well now all the AI stuff is broken. ACP sounds cool until you realize every single CLI in existence works better.
My wish is that Zed gets the core working correctly 100% of the time before moving on to expanding feature sets. For now I'm back in NeoVIM because it always works the first time....
https://github.com/zed-industries/zed/issues/38109
Hopefully soon I can give it another shot at full time usage.
My experience has been so different. Zed seems to always do the right thing for me when I concurrently edit files with other tools. Not doubting your experience or anything, but you must have a very different environment than me. Zed has been absolutely rock solid for the past year on my computer.
With the AI stuff, it feels like they invested a bit prematurely. When the Agentic editing demo came out (6 months, 10 months ago? It’s a blur), it felt right. Accepting and reviewing edits, live tracking ..etc., felt like pair programming. The ACP addition felt like a natural evolution .
With the continuous improvement in CLI tools and people’s experience with them, it feels like doing a live review or edit-by-edit approvals all feel like a drag. I personally have come to avoid using the IDE/Editor. I just kick up Claude code - plan mode, auto-accept edits. Once the session is done, switch to the editor and make necessary adjustments. I suspect people with Max subscription and “dangerously-skip-permissions” …etc won’t even care if an editor has AI integration or not.
yup, my pet peeve is there are are not way to disable line wrap. the setting that exist doesn't work and there's no way to actually disable it instead of just increasing the max characters (with set hard limit in the source code).
have a big docs or log,data file where you don't care for the rest of the line ? well too bad better have a spare editor.
this feel to me like it should should be a number #1 priority. "an editor need to nail the editing part".
https://github.com/zed-industries/zed/discussions/26344
on the positive side I do like that you can in-place edit the result of global search.
The thing about "basic" here is that it's subjective. It could be that these issues only happen on your machine or that the staff (or even most people) don't need what you're asking for. Of course they should try to fix them anyway, but their backlog is enormous.
Curious about the failure to detect FS changes made outside of Zed. Are you on Linux?
I am not getting tons of issues with Zed going out of sync all the time. I wonder if the issue is it silently having issues watching the filesystem due to open fd limits.
I've noticed that not only does it sync but it even will recognize if I rename the folder a workspace is in.
But then, I've run into a couple of strange issues that tell me there is more polish needed:
- Sometimes upon using LSP refactor, it seems like if a bunch of files get renamed, the open buffers will get screwed up somehow. Like, I'll hit save and it will write to the old filename! It's not actually a huge problem, as I can close the buffers, delete whatever excess files I accidentally create, and re-open them without error, but it is confusing as hell.
- I have indeed had issues with the file view not always updating when files are added externally, however it is not constantly. I usually just reload workspace when this happens. It is a minor frustration, but I had many minor recurring frustrations with both VS Code and Neovim before too, so I don't consider it a deal breaker.
Oh yeah I’ve noticed both the desync and AI not working via ssh.
Also buggy git support, I selected a few things but it committed everything and made me think I lost it all.
But I love Zed when it works. Literally 5h more M4 battery life vs Cursor.
Agreed. I hear way too much about Zed considering the editor doesn't allow a window to be popped onto a second monitor.
Yeah, I have been fighting Zed to get agents to use podman on my host, but Flatpak is sandboxed and makes it almost impossible. The ideal solution would be that Zed could use podman or docker to spin up a container where agents could run free!
Agreed, we should support this!
How does Neovim handle outside changes then? Or is there a plugin to make it work? AFAIK it doesn't reload any bufferes when files change. IntelliJ is the only other one I know that does it transparently.
In vanilla neovim you can use autoread... this does depend on a focus event like entering and leaving the pane or switching buffers. However, for my workflow which is "go to a different terminal pane and do some things then switch back" as soon as I focus the buffer it updates.
Where as with Zed it'll just keep showing the old content and in fact closing and opening file wont even change what it shows in Zed. It's really really annoying. I have to exit zed and open it again. This means if you are working with AI agents you end up having to do this often.
Plain vim asks what to do by default. A single 'l' does a reload.
You can configure Neovim to either auto reload or to ask you whenever a file is changed. (Can't remember which one is the default, probably to ask.)
vim.o.autoread (default on) When a file has been detected to have been changed outside of Vim and it has not been changed inside of Vim, automatically read it again.
I'm ~80% sure auto-reload is on by default, and 100% sure that it's on in a fresh LazyVim install.
Thanks, just found something about that.
I agree with you 100%. If basic functionalities are not airtight, there’s no way I am going to deal with growing pains just because they want to get paid on AI fluffs. Contrast this with something like Ghostty.
Zed looks really cool and I would love to give it a try, but I am just too beholden to devcontainers. I know there are workarounds to use them with Zed but with many extra steps compared to VS Code and it's forks. I can't go back to not having a totally integrated container per repo.
I understand interacting with other engineers right in the editor but I dont get why so many collab tools need to be bolted on to what is basically a text editor. This will only fragment communications since its not just engineers that work in any company. Meaning you'll now have communication spread out in Slack and Zed making collaboration difficult, not easy.
I dont honestly dont get the allure of pair programming. My pair programs are the unit tests which I run as often as I can and limit discussions to Gitlab/Slack. I have worked ag FAANGs and large companies and never once pair programmed anything.
I honestly cannot think of a single software or process problem that requires real time collab in the editor. Having said that it is a cool feature and I quite like Zed as an editor.
I _really really_ want to try this feature, but only if I can selfhost the collaboration server. If there is any way to do this, it's not obvious. Given that as I understand it, lots of project details will pass through Zed's servers, I can't imagine any enterprises would knowingly allow this without some kind of SLA with Zed.
It could be easier, but it is supported AFAIK: https://github.com/zed-industries/zed/issues/8260#issuecomme...
Unfortunately, we no longer support self hosting. We're planning on reintroducing it once we've polished the single player experience a bit more :)
Oh, that's unfortunate. Why not support it at all, even if just for people who are kinda hacking it in?
We’ve been growing and have had to scale authentication beyond what our original collab server could handle. Not many people are using collab yet, we’re prioritizing non-collaborative features (like our recent Windows release), and we’re planning on rebuilding all of this on top of DeltaDB. Fundamentally, it just fell through the cracks.
Self hosting will be a vital feature for users and enterprises though, we’re planning on revisiting it once we have a few more features settled :D
Looks like development of DeltaDB has been discontinued.
https://github.com/delta-db/deltadb?tab=readme-ov-file#-upda...
That is a different project :D
Here's our overall pitch: https://zed.dev/blog/sequoia-backs-zed#introducing-deltadb-o...
Oh is this based on my favoritest data structure ever: zed's Sum Tree?
Looks great. I use some CRDTs like Loro and Automerge, are there any learnings or collaboration you're taking from them to improve DeltaDB?
Also the name, I thought it was a database with CRDTs due to the "DB" at the end of DeltaDB, but it's a version control system, I thought that was somewhat misleading, any reason why it's named so?
Cool, thanks for the transparency and info.
I'm really rooting for you guys, and your direction and quality is exciting to see.
Thanks for the update. Can't wait to hear more!
I really really don't want comms or multiplayer tools in my IDE.
Don't bring the attention economy to my cave of solitude, it's where I go to escape all that noise
It's not something I am excited about, but it is something I want my IDE to do well if I must engage with it. Other remote pair programming experiences are even worse and I appreciate Zed's capability in the area even if it's not what I prefer.
A lot of my IDE choices are about extensibility and flexibility more than perfection for my preferred coding approach. After all, until I only work for myself I need to be ready to accommodate the needs of others as part of my job.
I removed the collaboration panel from my bottom bar and don’t have to deal with it at all. Can recommend
I stay away from anything Hashimoto is making after my experiences with TF & HCL, or at least the bar is much higher / projects get a lot more scrutiny
Yea you’re thinking of Ghostty, this topic is about Zed, which is not related. In fact Zed uses Alacritty for their terminal, I wish it was Ghostty!
Zed is sticking to all Rust, it appears, so Ghostty wouldn't work. Also Ghostty has a lot more UI features itself like tabs etc so I don't think it's as easy to integrate in an editor which also has its own UI, so Alacritty being mainly TUI driven makes more sense.
What does Mitchell Hashimoto have to do with Zed beyond a guest blog post?
My understanding is that he is involved in the development of Zed
I find Zed's feature choices to be really poor, and then I probably made a poor association based on his talking about zed
Hashimoto writes Ghostty in the Zig programming language. There might be a mixup here between Zig and Zed.
I know he's working on ghostty, but I've never heard anything about him working on Zed, was that what you were thinking of?
This feels like a huge distraction from building an IDE.
A monnumentous yak shave.
I've never used or cared for multiplayer in VSCode or JetBrains. It's silly.
I've never been the pair programmer type. The only time I've needed to share an IDE is during a SEV or ridiculously complicated systems bug, and that's 1% of the time.
Their multiple rounds of VC funding are predicated on their vision of collaboration so they gotta make a go at it.
I like Zed. I pay for pro. I like the integrated agent stuff (though my usage model has changed a bit after 5 months of use).
I'm happy that others can type in each others' space, but this post reveals a tension here. They are building a tool for building the tool, and their own team. I think that's cool, but at a 2-3 person shop heavy polyglotted across 4 OSes and 5+ programming languages, this is not what I really need.
What I'm looking for is a snappy tool (check) that lets me explore, understand, modify code at a next level (marginal). And I want it to not only be snappy by virtue of execution efficiency, but cognitive load. I want the less-is-more experience. I don't need it to do Swift, Kotlin, or Python, because there are bespoke IDEs for each of those that focus on the environments where I deploy them best. What I mostly want from Zed is the ability to see the outline panel at the same time as the directory panel, and to separate the search outline from the file structure outline. I spent too much time toggling views in Zed.
> What I mostly want from Zed is the ability to see the outline panel at the same time as the directory panel
You can do this now by moving one of them to the right dock (right-click the toggle-button)
And then toggle between that and the AI agent? Zed is a workbench that lets you put two, and only two, tools on your "workbench" to either side of your text workspace. I want to put more tools on my benchtop.
> And then toggle between that and the AI agent?
Yes
- Do you want to split a dock vertically?
- Do you want to open the panels inside an editor split?
- Do you want to detach the panels as separate windows? (https://github.com/zed-industries/zed/issues/17618)
Maybe I'm old, jaded, stubborn and paranoid, but something about a coding editor that is controlled by a company is off-putting to me. It's even more off-putting when you add Zoom, Slack and everything else into said editor.
…you’re free to use other editors? People like Zed. They like IntelliJ. They like VSCode. If you have an aesthetic preference against all professionally maintained IDEs, I think you’re in the minority.
...doesn't mean the majority is right :)
There is a lot of complaints about Zed in the comments here. I don't think that they are "hate", per se; they all definitely care about Zed and want it to succeed.
I daily drive Zed for work across several languages and I love it. I use a lot of its features, like the git interface, agentic editing, etc. I might even consider paying for Pro in the future if I want unlimited edit predictions.
However, all of these complaints are fully justified. I think Zed is a massive undertaking, only one that a VC-backed company has the capital to do. iirc, it requires 70k lines of Rust just for the cloud part [1]. I cannot fathom the amount of fundamental infrastructure they have to get the editor functional at all. That doesn't excuse all of the papercuts in Zed though.
If I were Zed I would do the following:
1. stop all work on future features, like DeltaDB etc. They all seem extremely cool but they won't meaningfully contribute to increasing Zed adoption or fixing its issues.
2. remove all agentic editing features. if Zed tries to simultaneously become the world's best agentic editor and a good general-purpose text editor, it will fail at both. Keep around ACP so users can still use other agents, but remove all of Zed's built in agent stuff.
3. fix literally every papercut. Triage every single issue and go through every PR, even if it will take half a year to do so. People won't switch to Zed until it's perfect, and the existence of this many issues means it's not perfect enough.
4. make extensions actually good. Every programming language, library, etc. has it's own ecosystem, and many such ecosystems mainly rely on VSCode extensions for advanced features. Zed needs to be extremely extensible like VSCode is; obviously its architecture makes this slightly harder, as it's nontrivial, for example, for extensions to render their own GUI, but there are a lot of low(er)-hanging fruit for extensions that need to get solved. People will only switch to Zed if they can get a similar breadth of ecosystems.
Of course, this won't happen, and given that none of these will really make them money, Zed has no incentive to focus on these, especially given the amount of time they would need to do this. But I think that if Zed can't nail the core experience, it won't get anywhere.
[1] https://maxdeviant.com/posts/2025/head-in-the-zed-cloud/
> Despite attempts to make Atom—an Electron application—more responsive, it never reached the performance standards the team yearned for.
This feels like an attempt at deflecting blame. VSCode is another Electron application that ended up having better performance than Atom. There's another Electron adjacent application that has good performance, the one you're probably using right now to read this page.
Depending on page content of course
What are you saying here? It is true that VS Code is less bad in terms of responsiveness in comparison to Atom. Zed, however, is written in Rust (i.e., not Electron), and I would guess it is at least an order of magnitude more responsive than VS Code across every possible scenario.
Web technologies are an unrivaled technological marvel for what they are, but it is disingenuous to imply they represent anything near the peak of what we are capable of in the context of performance.
It depends. It definitely opens faster and the general UI seems a bit faster, but open a largish file (a few MB) and VSCode will easily out-perform Zed because it doesn't have that fancy CDRT thing.
In my experience VSCode is plenty fast. Use it with no extensions and you will have zero problems with performance (though memory use isn't great). The real problems come when you have extensions, especially because it's often impossible to attribute performance issues to them because they can often do a lot of work all in the same "extension host" process.
Why is Zed using a CRDT when I open a source file to edit code? Are they using it for more than their multiplayer stuff? (agentic stuff?)
Based on this comment [0], they're building DeltaDB as a version control system which uses a CRDT, so I assume even in single player mode, the file will instantiate its own CRDT for fine grained tracking of changes.
[0] https://news.ycombinator.com/item?id=45916196#45919739
Zed is lovely and I hope it becomes super successful but this kind of mass collaboration might be ok for meeting minutes... maybe. But thinking of it for coding it gives me shingles. Code by mass live committee. Yikes.
I think it's a fun and interesting idea for training junior engineers and possibly for other use cases. Suggesting alternatives to (perceived) bad practices the instant you see them could be helpful for many people, and also save a lot of future time for reviewers.
I could also see it as a potential productivity aid. Person 1 sees Person 2 is writing something and they don't want to be seen as idle, so they start working as well. This might sound oppressive but a lot of people who struggle with ADHD/procrastination/akrasia actually receive great benefit from that structure. Similar to that startup that forces you to code while screensharing with a stranger in order to push you to work, or people who code in cafes/libraries to be more productive.
As long as it's not an organization requiring it for senior engineers, I could see promise to it as an eventual common new paradigm.
You can use remote screen control with Zoom to train up juniors. More often than not, you'll want to share the whole screen so you can show them webpages and other things that do not live in an IDE
Pair programming can be really great. Or horrible. Depends entirely on the people.
This would be good for code-walks too though. Instead of having to share your screen and hope the video comes through well. Everyone can follow along in the comfort of their own editor.
> code-walks
it's probably subjective, but I find these collaboration features can be overused for this kind of thing.
If someone is walking me through something, I just want to see what they see so I can focus entirely on what they're saying and no part of me is distracted by having to follow along or seeing other code.
I know typically these collab modes have an auto follow feature, but it's not as simple as just read only video being streamed to you, there's loads more ways it can go wrong and add noise / distraction that provides no benefit.
Problem is video is expensive and compression can get bad.
I agree being able to see the pointer is important, since not everyone is good about moving the cursor around.
> Code by mass live committee. Yikes.
Let's lean into the chaos and see what it might give us. Imagine a production application deployed directly from a non-version-controlled directory. Anyone on the team can edit the files, at any time. Insane? Probably. The disadvantages are easy to see.
But the positives are really compelling: 1. make small, granular testable changes; 2. use feature toggles; 3. refactor intensely and concurrently; 4. always work on the latest code; 5. use in-code documentation instead of GitHub/etc workflows; 6. explore continuous, incremental, hot-swappable code deployment.
Doesn't thought of ditching all the wasted motion and ceremony around logging async work and just coding sound glorious? I'm actually not a "move fast and break things person" usually. But the idea of moving so fast that broken things will only stay broken for a tiny fraction of the time is pretty compelling. There is also an intensity that comes from real-time interactions where a team needs to reach consensus quickly.
Feature Toggles: https://martinfowler.com/articles/feature-toggles.html
BEAM (Erlang, Elixir) provides hot-swappable code and lots more
It's just pair programming when you're doing it on code so if you can bear pair programming you'll be fine. Personally, I hate it.
Pair programming usually has a single "driver" on the keyboard to keep things controllable. Here, everybody is driving: "dozens of cursors are concurrently editing the same file in real-time."
The concept of sharing and taking turns has been lost on the software engineer here....
a few years ago our company used Screenhero which allowed editing with multiple cursors while screensharing.
The experience was actually quite nice for two-three people but we always had the "ok let me type now" flow. Multiple changes happening at once sounds hyper distracting.
my understanding is that this is the dynamic in modern college classrooms. everyone opens a big shared google doc for notes and they all collaboratively edit a set of notes in real-time.
or at least that's what i've heard, no idea if they actually do it.
it is nice to see a crdt backed editor tool for markdown and code though. gdocs markdown support has been lacking for years.
> or at least that's what i've heard, no idea if they actually do it.
Yeah, we used to do that back when I was in college. It's only for certain classes and most people usually kept their own notes too (or instead, why write twice). Or some classes ban laptops so you'd write on paper anyway.
True, but the option of wysiwyg for editing markdown would really be a great addition. Or even just for preview ... https://github.com/zed-industries/zed/issues/21717
That's not how they, or anyone else, uses it on code though - that's on their notes. This is just a feature, it's up to you how you use it.
The metaphorical infinite monkeys on typewriters.
Actively programming in pairs (or more) is also not for me. Reviewing work async is great IMO though.
Yeah, people are different, but a lot of this difference results from various constraints, such as cultural practices around collaboration or technological options. Many of these limitations probably shouldn't be locked in by our tools or norms.
When learning a new way of thinking or moving (i.e. martial arts) people often really benefit from high-bandwidth, low-latency, shared-viewport-onto-reality interactions. Watching someone's cursor move while they talk is one way to get a window into their problem-solving toolkit.
These are definitely some interesting features, though not sure I'm in any position to take advantage of them at all.
The multi-user editing is kind of cool... there's an ANSI art tool (PabloDraw) that you can run a host session so multiple artists can create text art, and I thought back when I first saw it, that it might be cool to be able for multiple editors to work on a project. I've used some of the collab stuff with VS Code, but haven't done enough to even begin to compare.
Not to mention that in a lot of workplaces, self-hosting or otherwise layers of bureaucracy stand in the way.
I would love to see collab servers take the same path as LSPs in being standarized and integrated across various editors and IDEs. I would love to work more closely with my VSCode peers, for example. Of course some features may be outside the standard and only supported with likewise editors, e.g. voice chat perhaps, but having shared cursors and a text chat would be a good start.
This is what I'd like to see as well. These collaboration tools are really good, but I barely use them because they always assume that you and your team are using the same editor. Most of the time that's just not the case, so I've used them a handful of times but beyond that there's little opportunity.
It's probably not an issue the Zed team will experience as they're all naturally using their own editor. Hopefully it's on their radar though.
> because they always assume that you and your team are using the same editor.
Network effects are probably a strength for a company, not a drawback (which it is for the user of course). Even VSCode has some notion of network effects, such as their proprietary extension store.
From reading some of the comments, I'm not sure which is worse, Zed Shaw or Zed the code editor
I use Zed instead of my normal text editor because it opens instantaneously. But I don't write code in it, and still use my IntelliJ + IdeaVim with Claude Code and Codex in a separate terminal in Ghostty.
But Zed is an insanely fast text editor to open text files in. Just double click and it's on the screen. Love that. Maybe over time I'll do more in it.
If that is all you need there has always been Vim and Nano.
I do use (and am in the Backers.md of) neovim. But the GUI versions open much more slowly than Zed. Zed is near instantaneous. I do often edit text in neovim by opening a terminal window and doing things, but the experience of browsing a large log file and filtering through it is much nicer in a GUI text editor, so if I ever double-click or open a JSON file I want that to happen instantly.
I can't stand Nano. I find it impossible to use because my brain is strongly vim-bound.
https://en.wikipedia.org/wiki/SubEthaEdit
https://tidbits.com/2003/06/30/apple-announces-design-awards...
People have been doing collaborative text editing since the 60s actually! See, The Mother Of All Demos[0], referenced in our first blog post[1] :D
I'd say CRDTs are also a big change. CRDTs make live collaboration much more robust for all parties involved, and they only started to reach maturity in the mid-late 2010s
[0] https://en.wikipedia.org/wiki/The_Mother_of_All_Demos
[1] https://zed.dev/blog/crdts
SubEthaEdit was a very inspiring software project for me. The fact that a small team could, in a few months, produce an amazing app that solved real problems and gained notoriety was amazing.
As time goes on it feels like much of the low hanging fruit opportunities in software is disappearing faster and faster. I'm also a fan of Zed and everything they're doing, but it's notable that shipping next-gen editor software takes a lot more developer effort now than it did in the 2000s.
> it feels like much of the low hanging fruit opportunities in software is disappearing faster and faster.
Yes I agree but so many things that might seem "done" (and in someways I think software/SaaS as an ecosystem is "done" compared to where we came from).
BUT - so many companies just bloat themselves and their products. I think the end of ZIRP is going to have an effect on that (more enshitification / rent seeking for sure) and I think there will be an opportunity to iterate and make copyware that doesn't take the higher development efforts.
We really need a winning electron alternative that is more resource friendly. That, IMO, will be a big game changer and I know there are lots of promising alternatives already.
> but it's notable that shipping next-gen editor software takes a lot more developer effort now than it did in the 2000s.
Yes, the scope increase is vast, due to more languages, more tooling, more features, higher expectations, and more competition.
I'd forgotten all about it but SubEthaEdit was such an amazing tech when we were using to collaborate internationally back in about '04. It went off my radar but I am glad to see its still available as a free app.
I love Zed’s minimal design language ... clean, restrained colors, low visual noise.
The screenshot below surprised me:
https://zed.dev/img/post/zed-is-our-office/this-week.webp
All the colorful avatars and the busy side/top panels feel out of character with the usual Zed aesthetics.
I tried the collaborative features to pair program with a colleague a few months ago, but it was bad. It was very flaky in establishing a connection. In the cases we were able to establish a connection, the voice chat would not work. We tried to make it work for a couple of days, and then we gave up. Has there been lots of work in the past few months on the collaborative features?
It would have been good to include a link to the collaboration docs https://zed.dev/docs/collaboration in the article. There were a lot of links in that article and a lot of assumptions that I knew how things worked. And I daily drive and like Zed, but I had so many questions.
I was very surprised to find a "forum" integrated in Zed when I first opened it. But to be honest, it is not something I ever felt the need for and overall I don't like having this in my text editor. So far it never got in my way and that's a good thing, I hope it stays that way :)
Is this the new Zawinski's Law? Instead of extending to read email, Zed extends to enable chat and voice-video. :)
I’ve been using “The Notioning” for the last few years to refer to the convergence of tools like slack adding notion like features, clickup adding notion and slack type features, and so on. There seems to be a stable set of features that retains teams in an org
Perhaps we can call it the "Hashimoto Valley", analogous to the Uncanny Valley, but for DX
What does this have to do with Hashimoto?
HCL / TF DX and my missassociation of Zed / Hashimoto (see where I was corrected elsewhere in these comments)
Looks a lot like Google Wave. This is interesting for some things but I don't think coding is it, for the same reason that IntelliJ's CodeWithMe doesn't work for pair programming. And apparently one of the guys is a former Pivot, so it's a little surprising.
Pair programming is Two People One Cursor. A critical aspect of it is you're both looking at the same lines of code and working on the same problem and following each other's thought processes.
CodeWithMe (and it seems Zed) is Same Codebase, Same Day. There's no shared focus. You edit stuff, I edit stuff, maybe there's overlap. But this isn't much different from doing separate git commits.
So far the only remote pairing tool I've found that works competently is pop.com.
With zed you can also share your screen in the editor which makes it a bit better, but still you can't take control of the other machine.
IMO if you only care about coding doing it in the editor is the best approach, you get zero latency and have all the context that you need (most of the times). But if you want to do more, like opening the browser for whatever reason, or teaching how to use a specific cli, etc, then taking control works better.
If you liked pop you might like gethopp.app, which is an OSS pair programming app (full disclosure I am the co-maintainer). Unfortunately because we have chosen tauri for the frontend we can only support macos and windows, but I am working on a solution for Linux too.
Pair programming so often degrades into one pilot and one person just sitting there trying to catch mistakes. When those mistakes are caught they are mentioned taking the first persons attention away and breaking flow.
In contrast how I like to work, with similar level people, is to work at the same feature in the same codebase at the same time. We either sit next to eachother, or have a remote call, where we continuously talk through what we want to achieve. Sometimes this results in one person writing ahead (code, docs, doesn’t matter) and the second sweeping behind it and cleaning it up. Two cursors, one source. IntelliJ even manages to keep authors correctly in git.
The other mechanism is where we work on different parts of the code base at the same time. Either main code and tests, or split across interfaces and implementations. Because this happens on the same machine the iterations are way faster as they are local and incremental.
This basically saves the whole dance of creating branches, pulling/pushing, the fixing typos etc.
/offtopic
Is it just my vision, or are websites getting super low contrast these days, esp the text-heavy ones?
I feel like that's been a trend for the past decade at this point. I don't think this one is particularly egregious but it ain't great either.
Could be your monitor as well.
Also the typeface, kerning and line heights are a bit - unconventional, which doesn't help the readability
I'd say it's medium gray on white, which is not too bad in my subjective opinion. I have seen far worse. Light gray on white was "trendy" for a while and dark gray or dark green on black has always been popular among the edgy crowd.
Any plans to leverage 3D at all in the interface?
was floored by the "explode all layers in the user interface and simulate a 3D camera rotating around them" graphic when i first saw it !
3D is always difficult to get right, but felt it had some really cute possibilities,
any way to open this up so devs can try things out?? < 3
> was floored by the "explode all layers in the user interface and simulate a 3D camera rotating around them" graphic when i first saw it !
What is this in reference to?
https://zed.dev/blog/videogame !
thanks!
i'm now guessing the software engineering universe is becoming like the lawyer ones
1. big corporate shops / vc funded ones - many tens of programmers working on features (this is where zed collab features might be needed) 2. bespoke high productive small teams - less than 5 product programmers in a company e.g basecamp - these would be your bespoke law firms 3. the indies (injury lawyers) -> 1 - 3 programmers chunning out products at scale or eating of one product + maybe with help of A.I
for 2 & 3 - a lot of stuff being shilled is not needed. a legal pad + some notes that can be posted via a google doc is all that's needed. Jira isn't needed too
I ... don't like this one bit. I hope Slack doesn't start including a text editor.
There are canvases.... Some similarities if you squint. Clearly not for code use but for shared durable notes....
To me it feels like 2 different set of products, what they are showcasing here seems very similar to slack/teams.
Of course, it would be awesome to have a faster and open source slack, and if I can take notes on the same style as my editor great. So I guess, it would be nice to be able to embed zed in another product.
I think this would be appealing for a company that it's core product is code, like zed, but I do wonder if other companies even need this functionality.
Please work on decreasing the binary size - it's whopping 400mb!
Technically really impressive. In practice, completely unpractical in any medium to large organization. And although I adore Zed's speed and reliability, I still don't understand why we need these features at all.
Because Zed is not somebody's side project, it's a business looking for ways to pay everybody's salary
> Collaboration as it stands today is considered alpha, and for the time being, is free for all to use! Peruse the source code.
Because "maximum editor" was achieved with Visual Studio Code (which is why all these new editors look and feel like VSC), the same way "maximum toothbrush" was achieved with the electric toothbrush. But the toothbrush industry had to keep going, so now we have a $400 bluetooth connected toothbrush-as-a-service that monitors your brushing habits to optimize teeth cleanliness, with brush heads that cost more than 5 regular toothbrushes.
Zed and the current crop of AI editors (including VSC itself) are that toothbrush.
Here is where I've settled on for Zed. I initially thought it might be a Sublime replacement for one-off files, but it seems it's geared towards projects. It's not as powerful as Jetbrains (RustRover, PyCharm etc), but is much faster. So here's how I'm using Zed:
I think its' a much nicer experience than VsCode, which I admittedly haven't figured out to run in a project-oriented way.I'm also trying their GPUI library, but am in the early stages, so can't really comment on how it compares to EGUI.
Zed is the only modern IDE-like editor which is fast enough to replace (n)vim for me. I plan to use it for more and more projects, but I've had minor issues with it's Vi-mode.
I'll always remain someone plugged into vim because I need it sometimes when shelled over a terminal. Editing files over SSH can work with editor support, but is often less reliable or fast than jumping through whatever hoops I need to to get an SSH connection once and then doing everything from there.
Incidentally, I use Vim for editing files via SSH as you do, or if I'm in WSL, but haven't figured out how to use it for projects!
I just have a few plugins which help. Mainly the LSP for gotodef and popovers for type info, etc. This was what finally made me transition to neovim. Also a tree viewer, Startify, and :Rg for ripgrep integration. Those are my big ones.
Sadly my workflow of using `!` to get back to my terminal and things like `!make` or `!cargo build` is fucked in neovim. So I do a lot of ctrl-z and the a lot of killing stopped processes I forgot I suspended. I've complained about this in various threads and chats, but the developers aren't interested in letting us use the old vim `!` which is super lame.
> Sadly my workflow of using `!` to get back to my terminal and things like `!make` or `!cargo build` is fucked in neovim. So I do a lot of ctrl-z and the a lot of killing stopped processes I forgot I suspended. I've complained about this in various threads and chats, but the developers aren't interested in letting us use the old vim `!` which is super lame.
I'm sorry... what?? I still use vim as I haven't found a reason to jump to neovim, but you're telling me external commands don't work properly? That's wild.
They "work"... but instead of switching back from the alt screen and giving you your terminal back, they open a gimped little window within vim which makes it hard to navigate, copy, and search for things in the output.
Why would you use a tablet for this type of work? Honest question.
Surface tablet, so it's like a laptop. I can kick the keyboard off when I'm reading stuff, to allow room for food and drinks. And I do drawing/notes with the pen.
Another way of stating this: It's a general purpose portable computer; not specialized coding PC.
I lost all faith and interest in Zed after they introduced AI features.
I feel the opposite way, but fwiw you can turn off all AI features in Zed by adding `"disable_ai": true` to your settings.json.
The AI features work well but to each their own
That's alright, there's always Sublime Text.
Or TextMate -- that Sublime was, let's say, "inspired by."
you dropped this king
Why? Its still incredibly plug and play, by default you don't even see it
Haha it's like Google Wave!
Very cool idea, and helps promote owning your own data, and it being highly interoperable (plain text!)
I do wonder if we need a term for shoe-horned dogfooding though. Like sure, you can do this. You could do this in Figma! Or in Notion! Or in LEETCODE if you wanted to.
At least with Zed though, its plain text. If you find another way to collab realtime on plain text, you're not bound to 1 vendor.
You guys need to figure out how to create Slack shared channels in Zed and we're all switching until they won't be needed anymore.
Can you share more on this.
While I do not work at Zed, I'm curious to hear more about this use case for my own company needs.
Your company has a user pool, you sign a BAA or start working with a partner company that has their user pool. Instead of creating slack accounts in both you can share external slack rooms that only people that are invited in/from their respective orgs can join without having to co-mingle employee user pools.
But why would external partners want to look at your code? I guess if you're also integrating with them? But generally you just give them repo access instead. For Slack, it's different as messaging is a core feature to collaborate between different people in different companies, but looking at code is a very specific use case.
I hadn't realized Zed was built from the ground up to support collaborative programming. I liked it already, and I like it even more now.
I could imagine that in ten years git will feel strangely slow and ceremonial. Why not just continuously work and continuously deploy live-edited software
I feel the opposite way, that git branching and merging will become a bigger part of the job as more code is written by agents in parallel and then accepted by other agents or humans.
for now yes absolutely. but I’m already hearing rumblings that some people are having luck letting multiple agents edit the same directory simultaneously instead of putting changes through PR merge hell. It just needs coordinations tools, see https://github.com/Dicklesworthstone/mcp_agent_mail as one (possibly insane) prototype
for example it’s not out of the question that we could end up with tooling that does truly continuous testing and integration, automatically finding known-good deployments among a continuously edited multiplayer codebase
we’d have to spend a lot more energy on specifications and acceptance testing, rather than review, but I think that’s inevitable - code review can’t keep up with how fast code gets written now
Having tried a janky version of this myself with a NOTES directory, I am very bearish on this being a better workflow than just improving the ui wrapper around git worktrees and the isolation that provides.
Codex already has a fantastic review mode, and gemini / claude are building tools around pr review that work no matter how that pr was produced, so I think this interface is going to get baked in to how agents work in the near term.
That’s a very optimistic outlook for the future.
Often projects need a history of stable checkpoints, and source control is one way to provide that.
Yes, but does it need all the ceremony surrounding it? If, every time I saved the file, the changes were analyzed and committed to git, and a useful commit message included, and commits squashed automatically and pushed and tested and tagged (using magic, let's say); if the system existed in the background, seamlessly, how would our interactions with source control and with other developers look?
automated commit message will tell you the "what" not the "why".
In any circle of "what makes a good commit message and why even do it" discussions, invariably the recommendation is to explain the "why" and leave out the self-evident "what".
If your stance is that commit and commit messages can be automated away then we might as well not even have them.
I don't share this view, but yeah in this world we don't need AI to do things that shouldn't be done in the first place.
increasingly, the automated systems have access to the original ticket or bug report, and maybe even the conversation while implementation is happening. They can record the “why”
Use jujutsu
Counter argument to living software is that it treats "never done" products as a virtue instead of a failure of design.
Here's a thread where the person replying to me makes this case: https://news.ycombinator.com/item?id=45455963
I love it when I have a tool that’s “done” but the software I work on in my career is never, ever done. It’s almost like there’s two different things we call “software”. there are tools like, idk, “curl” where you can use and old version and be happy. and there are interactive organizations in the world, like, eg, Hacker News, which mutates as the community’s needs change
Software for evolving business-needs is the same for me. What's insightful is that we (I) take continuously evolving software as just that: evolving. It's a defacto virtue to continuously tinker.
Doing away with check-ins entirely is the extreme end-game of that pov. I'm in product and every day and every week yes we very much continually change the product!
But I'm growing less convinced that the natural end-state of this methodology produces obviously better results.
it doesn't work quite well for complex projects that require integration with other teams/software.
You would need to either have separate versions running at the same time or never do breaking changes or devise some other approach that makes it possible.
It's not always feasible to do it this way
I think that’s a tooling problem. Maybe we do end up running a lot more versions of things in the future. If we believe that code has gotten cheaper, it should be easier to do so.
I wonder how many nines of uptime your team is required to have..
Imagine if someone clicks the deploy button when you're in the middle of typing something and then the service goes down due to a syntax error. To prevent this, we will need some sort of way to set a global lock to indicate that "I'm not done typing yet" and you can only deploy once everyone has released this lock.
Or you don't deploy unless it makes it through at least testing, and a build started while someone was editing the code would probably fail fast unless you coincidentally hit the button right when it's valid, but wrong, code.
I have been trying to figure out how this works in concert with Git (or SCM in general). Is one of the developers in the session merely responsible for it?
I mean, you have the same "problem" when peer coding in person. Whoever is officially working on the fix will commit it. I've helped devs get around a hump for ages, you don't get "credit" for all the work you do. It's why I hate most ticketing systems (when management starts to ask why your tasks fell behind), they don't let you correctly track multiple people when they work together.
Does a `Co-authored-by:` trailer in the commit message not resolve this issue?
commence feature creep
The collab editing stuff was Zed's original gimmick. The AI stuff is the real feature creep.
Without AI, the company would be dead in the water. You can very easily disable all AI features.
I think Tuple is a better collab app, but far more expensive.
The collaboration tools built into Zed have basically existed since the product launched. It is one of the primary drivers behind the product - they wanted to build the best editor for remote code collaboration.
I think the most difficult question are going to be how do you constraint that core feature without ever wanting to add more to it?
For instance, collaboration is a huge topic. You can have coding collaboration on the file, and that would be basic and appropriate, you can then replicate slack and you'll have chat rooms, which is entering creep territory, but it's natural! Then soon the chat room will need to link with issues and you can now have TODOs linked to some kanban board and we should be able to speak while we code on the same file! And this goes on and on.
It's exceedingly rare that the organization found hard courage to specifically avoid features that looks like easy pickings for the purpose of avoiding them.
Many of these features were in Zed since first release.
Also Zed was announced as a closed source comercial tool.
This looks more like a collab note-taking app. Don't know about code since i don't code anymore inside an editor but for collab things who knows
Don't tell the Posthog guys about this. Far too much collaboration going on here!!!
Haha: Can't find a good Windows laptop.
It's true, most of them are bad. Galaxy Book5 Pro or Microsoft Surface are OK.
Ever since Apple Silicon macs came out, it's been a real struggle. Almost 6 years later and there is still nothing on the market that:
* Has the same level of performance
* With the same or better battery life
* With the same quality of screen
* With the same quality of speakers and touchpad
* Runs as quiet or as cool
as the Apple Silicon macbooks. If you add in "needs to be able to run Linux" your choices go down from maybe 1 or 2 to 0.
They all have some sort of compromise. Either the speakers, screen, keyboard, touchpad, build quality, battery life, or thermals.
I have a Surface Laptop 7 with the Snapdragon X Elite, and it's pretty close. Checks the boxes for Screen, build quality, and touchpad. Loses out on speakers and battery life, and the fans need to run a lot more than my M4 Pro MBP does. It also loses on performance, and it doesn't run Linux. Windows on Arm also still has a lot of little quirks and bugs that start to become daily annoyances.
It's incredibly frustrating. I want, essentially, my 14" M4 MacBook Pro, but in a Linux laptop, and there's no OEM out there that's fulfilling that need without compromises.
Apple keeps pulling ahead in silicon and every other laptop OEM is just being left in the dust, shrugging their shoulders, and putting out the same old 1200p 16:9 plastic garbage they have always been putting out.
As long as I don't have to use, feel free to include it. It is really not essential feature for editor.
I run update and Collab requires you to sign in... which again, it is fine if you want it. I don't, so it can be dormant, icon is really tiny, doesn't take much space.
The feature of Zed that is most annoying yet essential is frequent updates. Pretty much daily when I switch to Zed window, I can expect update and restart, which messes up my window layout, so this is annoyance. Getting updates and knowing you guys are shipping good stuff is what is essential.
I think integrating terminal ai's is great move and useful. Sometimes I use it like that, often I use it in terminal (like the outside of the editor terminal) and switch to editor to review or update stuff. Same with git. I am old-fashioned.
Whenever a product tries to be too many things, it dilutes the core USP. Try to be an excellent code editor. Add extensibility. Done.
I get it, you are VC funded, investors want to turn this into a multi billion dollar unicorn.
Do not focus on investors, but developers.
Looks very cool, and of course it's nice to basically have Slack inside of Zed.
But personally, what I want in a Code Editor / IDE, is to be the very best experience at writing code and working with code projects. That is what will save me time and make the coding experience better.
Collaborative features are nice but not essential since there are other tools out there. It's not likely to move a team away from Slack (though if it's self-hosted, it stands a chance).
I'm not yet at the point where I can rely solely on Zed for python coding. I mostly use Zed because I like new initiatives, especially open source ones, and it's fast and responsive. But PyCharm is still better for python development at this point, with its one black mark being endless indexing on large codebases / dependencies, and I find myself falling back to it regularly. I would argue that the priority should be to achieve parity as a _code editor / IDE_, and then we can talk about other shiny new features.
Zed's dead, baby, Zed's dead
I love Zed. I, mostly, love the direction they are taking the editor in.
But. There are now two times I see Zed going in the wrong direction. The AI integration was one. This feels like the wrong direction again.
I never really liked the AI integration. It felt off to me. I do love coding with Claude and I think I know why. It presents the "information I need to know" in a way my puny brain can handle it. Colored diffs. Summaries of what happened. It isn't perfect, but it has been incredibly productive for me. I never got that from Zed's AI integration; perhaps this has been improved, but I was up and running with Claude in a way that I never was with Zed.
This write-up sounds like "slack in my editor." If it is that, I hate it. Slack has destroyed company culture and communication. People, who are inherently lazy (I'm an old Perl programmer, so I can say that), have stopped thinking carefully and writing carefully, and in that void just throw the first thing in their head into a slack channel and think that is "collaboration" and "communication." It's toxic.
For example, this comment rubs me the wrong way: "Staff members hop in, volunteer to show off a cool feature or bug fix they worked on, and get real-time feedback from the rest of the team." I don't think our human brains work well with "real time feedback" UNLESS we have the information presented in a way that gives us massive clues on what's right and what's wrong. Reading a wall of text is not the way. A colorized git diff, or a video, or an entirely new way of presenting information might make real time feedback possible, but I am highly skeptical a text editor is the way or place to do that. And, I'm an emacs user and love text UIs, don't get me wrong.
Do I want to have "generalized one off rooms for things that don't fit anywhere?" I definitely don't want that. I want you AS THE AUTHOR to be really intentional about what's important and fit that into the proper channels. I need to know that information, but I don't want to know about, nor have the unspoken expectation that I SHOULD have known, about the other stuff. And, I want "managers" (if that still exists) to be carefully thinking about those channels and how the company is organized and push that structure down to people in the organization.
As Zed is the office, having one off rooms instead of in person coffee time feels very dangerous. That's the world a lot of people live in, but I don't like that office.
If this comment is the guiding light, then I'm worried: "We're building toward a future where collaboration is continuous conversation, not discrete commits—where every discussion, edit, and insight remains linked to the code as it evolves, accessible to both teammates and AI agents." I'm human, I have kids, I have other interests. A continuous conversation is impossible for me. I want discrete ideas, and right now, discrete commits and PRs are better, IMHO, than what I hear here. It's hard, but setting the expectation that to be successful I need to be paying attention to a river of information flowing by seems like a bad idea to me. I don't buy that Zed solves the problem of hiding the pieces of information that I don't need to see.
Oh hey! I have an idea. Why not use AI to summarize those conversations into discrete pieces! </joke>
I do love Zed. It is the best GUI editor out there. I know they will get it right. I just am skeptical about this direction and feel it misses the forest for the trees.
Man, Im like the total opposite in terms of preferring the Zed UI vs claude code. I really try to avoid raw claude when possible. I very rarely pull it up to do concurrent sessions when I have Zed open already working on something else. Or if I need to do something quick while in the CLI in a random directory. Otherwise, I think just the "files modified" feature is worth using Zed as the primary interface.
You make some good points, and I need to revisit Zed+AI to see where things are at. This probably proves you are a better developer than me.
But, also, after reading your comments, I'm just not sure I need an "editor" anymore. I love that I can npm install claude anywhere. Zed does not exist for ARM servers yet, but I can install claude there, and it can troubleshoot my database connections, and edit code, and grep files. Those are all the things I used an editor for, because an editor has better ergonomics than using the CLI. I'm sad to say "misspelled prompts" might have better ergonomics for me.
If I were still hardcore avoiding a GUI Id probably be in the same camp. But I moved to VSCode as my primary interface a long time ago, and Zed is just a better version of my workflow when comparing against that.
100%. I also moved to VSCode years ago. Then, I got disillusioned by the performance, and by M$ telemetry. Zed is so much better in so many ways if you have to go with a GUI for all those reasons.
Just another fence for monkeys :)
Don't want to sound negative, yet when I read "it's in our DNA", I immediately lose interest.