Are we building too fast to think critically?
I recently read a great article about how Linear approaches design and AI https://designerfund.substack.com/p/ai-design-linear, and it got me thinking.
With AI making it so easy to turn ideas into working features, I wonder if we're sometimes skipping one of the most important parts of product building: questioning whether those features should exist at all.
I've definitely caught myself doing this. I'm building a travel app as a hobby, and sometimes I implement something simply because I want to see if it's possible. Then I realize it doesn't really solve anything and end up removing it.
Of course, experimenting is part of the fun, and building can be a way of learning too. But I really admire Linear's obsession with craft and thoughtful product decisions, and I wonder how we preserve that as building gets faster and faster.
How do you balance the freedom to build almost anything with the discipline to build only what matters?
Replies
I did this at a bigger scale. I built paid plans for LLMPvP, my arena where AI agents play chess and Go, before it had a single user: an analytics dashboard, move-by-move review, a cap on agents per plan. When I finished, the accounts table was empty. Billing is still off and every new account gets the top plan for free. What helps now is a roadmap section called "explicitly deferred": payments, tournaments and user roles sit there, each with its reason, like "no proven demand yet".
@wwenioaguiar That's a great example! I really like the idea of an "explicitly deferred" section in the roadmap. Looking back, would you approach things differently now? Maybe focus more on getting users and validating demand before building those features? Is that what you're focusing on now?
@claudiatd Yes. I'd ship the free version, put a waitlist button where the paid plans are, and only build them once people clicked it. The plans took days to build and nobody has seen them yet. That's the focus now: getting agents to play. I published the MCP server to npm and the official MCP registry, and I'm posting about the project here and on X.
@wwenioaguiar That makes a lot of sense! It's so easy to get carried away building things, especially with how easy these tools make it now. Thanks for sharing your experience, and good luck with the project! :)