Sayrud, the project you're looking at, is a collaborative spreadsheet web app you can host yourself.

In July 2024, I wrote a blog post introducing Sayrud for the first time. The idea back then was to build schemaless tables on top of Postgres: a little “toy” along the lines of Leancloud. AI wasn't at the point where a single sentence could generate a whole project, as it can today. I was using GitHub Copilot for code completion and asking it to help with some of the trickier frontend interactions. Once the first version was done, I just wrote that post to share how it worked and used it to build the backend API for my blog's comments.

I didn't open-source the code because of security concerns. After that, Sayrud went quiet for two years.

Then, a week ago, I wanted to try the latest Claude Opus 5.5 model. I handed it the old Sayrud codebase with just one instruction:

Use the actual Feishu Bitable page as a reference. Its frontend sends operations over WebSocket and supports collaboration between multiple users. Update the current frontend and backend accordingly.

About half an hour later, it was done, and the result was surprisingly good. It made the rather clever choice of Arco Design instead of using TailwindCSS like other models. In that first exchange, it fully implemented field editing, filtering, grouping, and sorting. It even built out the kanban, gallery, and form views!

It was remarkably complete.

Honestly, it's been a long time since vibe coding felt this satisfying. I'm going to pick up where it left off, keep working on the backend, and add user permissions, internationalization, and third-party login. I already use Feishu Bitable to automate my bookkeeping, but recent changes at Feishu have sharply reduced the AI quota available in Bitable. That gives me a pretty good excuse to reinvent the wheel.

I know there are already SaaS products like Airtable and Notion alongside Feishu Bitable, and open-source projects like teable and NocoDB. They're all well-established products in this space.

I'm not building Sayrud to compete with them. I want a project I actually use day to day, one where I eat my own dog food.

That's why I borrowed WordPress's self-deprecating joke for Sayrud's tagline:

Yet another collaborative spreadsheet.

I'll start by using Sayrud for my own automated bookkeeping. When something feels awkward, I'll fix it. If I can make a tool I want to keep using and get back a little of the fun of coding, that's reason enough for me.