blog / screen.garden is shutting down on September 1, 2026

Timeline of Key Dates

• July 1: Shutdown announced. Team (and thus subscription) creation disabled.
• July 15: Team expansion disabled: you may no longer add new members to your teams.
• August 1: All subscriptions paused. You no longer pay for screen.garden.
• September 1: screen.garden shuts down. All data stored on servers is deleted. Your local Obsidian notes are untouched.
• The Future: The source for screen.garden client and server is released and available for self-hosting and hacking.

Our plan with this timeline is to give you time to sync all your notes to a local machine for safekeeping, to add whoever you need to add in the next couple of weeks to do so, and then not to bother (or charge) you afterward.

Why shut down?

Many reasons. In the end it comes down to a single root cause. We had to grapple with this problem every time we made a decision, offered support, or built something new, and in the end we decided it was simply insurmountable with our present approach.

We couldn’t build our business on Obsidian. There are many facets to this problem:

Obsidian is closed-source software that we are fundamentally beholden to. For us this especially meant wrestling with editor integration; nearly every time a user reported an issue with sync, a true fix was unavailable to us because we can’t touch the interface between files and the Obsidian editor library, CodeMirror. We can configure CodeMirror, and did extensively, but we can’t control the lifecycle of a note at the level we need to, and never will.

While we always saw ourselves as competing with Google Docs and still very much believe in the vision of a collaborative document editor where you own your own data, in the end we were competing against Obsidian Sync. Obsidian is, for all its popularity, still fairly niche, and its userbase is not wholesale trying to solve collaboration in the way we expected. This meant that we were more viable as a pure sync solution with some nice collab features built in. And it is more or less impossible to supplant a first-party service with a third-party one, something we weren’t even really interested in doing in the first place.

And, finally, the fire has gone out of us both when it comes to being Obsidian users. Neither of us is using it on a regular basis right now. Every time I fire it up I notice little niggling s.g. bugs that are unfixable without Obsidian offering a more rigorous note lifecycle API. If you’ve ever written code you know how maddening such a feeling is. It made simply using Obsidian difficult for me.

Speaking personally, the rise of LLM integration in the Obsidian community also poisoned the well. I have zero interest in allowing a bag of weights to think for me. Our support load shifted toward supporting agents, our feature requests tilted toward MCPs, and the fire died. The degree to which the Obsidian community has embraced a corrosive technology that opposes—on a fundamental, structural level!—the kind of work that Obsidian was created to foster breaks my heart a little.

Which is all to say that we don’t have the market or motivation to keep pushing screen.garden forward. It’s an incredible piece of technology, we’re so proud of it, and the best path forward would be to give it to you for free—more on that in “What’s next” below.

We’re immensely proud of screen.garden

Before we get into life after screen.garden I wanted to write a brief paean to it. Sloane and I built a thing that works. We’ve had literally zero downtime since launch. It’s been running on the same slim AWS setup despite hosting way more teams and data than it did when we were testing it, just ourselves, years ago, and we’re nowhere near the ceiling for that box. It led us to speak at a conference on the backing tech on how we built a reliable system on top of unreliable data.

None of this would have been possible without a few key pieces of technology:

• The entire backend was written in Elixir and Phoenix, and we’d choose them again in a heartbeat.
• We used yjs for our backing CRDTs, which at the time was the correct decision. We wrote a custom Rust NIF to interface with it. These days we’d probably choose Automerge instead.
• We used Oban for a job queue and Broadway for processing AWS events for file sync.

The Elixir ecosystem, although small, makes this kind of work so much easier than it would have been in something like TypeScript, Go, or Python. Being able to diagnose issues live via a connected iex session and get live dashboards of the system for free is just incredible.

And, with regards to our own work…we built genuinely interesting software! Our architecture allowed for the server to remain the source of truth for notes without getting in the way of collab throughput. We could connect side effect to notes easily, collapse versions, all sorts of fun stuff. We did a good job, and it’s a little sad to say goodbye.

What’s next?

The key dates are outlined at the top of this post. After that…

screen.garden was never a full-time job for either of us. We will continue to work in tech. We will almost certainly continue to tinker at the edges of this exact problem. I know I have some ideas.

We plan on open-sourcing both the server and Obsidian plugin—all our code—under a permissive license. The server currently has a hard requirement on AWS to enable file sync, and we may put some work into easing that requirement. Once we open-source it we encourage you to play with the code and self-host it, but we can’t offer support for the code in the way we did for the commercial project.

Finally: thank you. And stickers!

I wanted to end by thanking everyone who’s tried, used, and given feedback on screen.garden since we first launched in 2023. We’re both still kind of in shock that people paid us for our labor on a thing we loved to build, and moreover that we broke even on it. Trusting us with your data was and is incredibly humbling. We hope you continue to collaborate, and are of course deeply sorry if we interrupt any core workflows for our users—we really tried to keep things going for as long as we could.

In a kind of ironic idiocy I purchased several thousand screen.garden stickers a few months ago. They have sat on a bookshelf since. If you’d like a few of these stickers please email me a mailing address. I’ll pay for postage, and I’d love to get these out into the world.

It has been such a joy to work on this thing together and to work with you all in building it.

In collaboration,

Kevin and Sloane.