Why we built Seamcult this way: a product design FAQ
I’ve written a few posts about the thinking behind Seamcult: why MCP, why there’s no built-in chat, and why it’s free during early access. I thought I’d bring them together here as a short FAQ, with quick answers below and links to the full posts.
Why MCP, and why no built-in chat?
You already have an assistant with your preferred model, conversation history, and a chat interface you know. MCP lets that assistant work directly with Seamcult’s drafting tools while you review the result on the canvas. This lets us focus on the pattern editor, and avoids charging you for another layer of AI usage through our own chat.
More on why Seamcult has no chat of its own →
Why does the model you connect matter?
Seamcult provides the tools and checks; the connected model does the drafting. In our experience, stronger models handle drafting noticeably better. Smaller models are still useful for tasks like arranging pieces, renaming, and exporting. That dependency is part of the product experience, so I want to be upfront about it.
Why is it free?
It’s free during early access because we’re still learning who gets the most value from it and what they need. A home sewist, a small studio, and a factory may want very different things. We’ll need a business model eventually. For now, actual use and feedback help us understand what’s worth charging for.
Happy to answer questions about any of these decisions, hear objections, or compare notes with other builders.
One thing I’d love your take on: would you prefer to bring your existing AI assistant into a specialized tool, or have the whole experience inside one app? What would make the difference for you?
Replies
I'd bring my own assistant if I can undo its edits on the canvas. If it gets a sleeve wrong, I'd want to restore just that piece and keep the adjustments I've already made by hand.
@al_kub, thanks for the reply! Yeah, we've tested a lot of agents by now and seen all sorts of things. They do make mistakes, so undo/redo is part of the architecture, though it hasn't taken its final shape yet.
Today the agent can undo and redo its own edits (and you can ask it to do that in the chat); a button for you is coming. The next big feature we're aiming for is full-fledged collaborative editing, so you can undo, redo or edit anything done by the agent, by you, or by your colleagues.
Feedback like yours is exactly what we're looking for to steer the product in the right direction.