CHANGELOG

The future of work is an arena for humans and agents to collaborate. Maybe this is a communication interface, like a chat layer, that everybody is already building. But somehow, I don't think so. Chat is an interface that adds a layer between thought and action. Humans are fundamentally visual and kinesthetic people, and my bet is that boards, canvases, and non-corporate tables, while still a layer between thought and action, will remove friction from doing great work in the same way that a picture is worth a thousand words. This isn't released yet, and I'm not sure when it will be. But it's what I'm working on in my free time.

I added role tabs, and ⌘ 1 2 3 to switch between them. I also pushed a few other changes to make the user experience better just based on talking to people. I also saw an opportunity for Pan to get really, really good by feeding it more context. So now, Pan has full context of who each team member is, so it can integrate as a leader and understand who gets what done. My realization here is that the value-add is if we can remove all the stress from what and where the great work is, so you can execute on the how and why.

The really fun part about getting feedback from users is forcing yourself to say "yes." I think great design is revealed by talking to people. Design happens not on the page you create, but in the mind of the user. If it's not immediately obvious, it's not good. It's like the push/pull doors. You can make the experience better by adding a sign to every door explaining how it works. Or you could just change the handle shape depending on push/pull.

Some stuff was just really broken. So I pulled my computer out and fixed it.

I like to consider myself a design-first person. Not in the sense that it should look good, but it at least needs to feel good. It wasn't really either of those. So I overhauled the design by observing and writing down anything that didn't feel immediately intuitive, and asking my friends to do the same test. It's not perfect, but I'm continuing to work on it.

I built Pandan with just one workspace, because it was only the small team at Quo using it. But then other people wanted to try it, so we scoped the workspaces. This was a big task, but it elevated the Pandan experience.

Pretty simple. We wanted to make Pandan more useful to the team. The biggest problem is that project management isn't in a place that makes it easy to collaborate. Managing the project shouldn't feel like work.

In 2026, is it really a good idea to make something that's not AI-native? So I spun up Pan, an AI agent to help you do your best work. The goal was to integrate Pan deeply within the architecture so it would actually be useful. We'll see how people use it.

Released Pandan as a simple Kanban board and task-tracking sheet. One evening on the ride home, my co-founder Jenny and I lamented on how there wasn't an easy way to "give each other tasks" that wasn't either extremely unintuitive or overly corporate. That night, I hacked out Pandan and unveiled it to the world.