ngl were all getting stuck in this
chat-only loop just because llms are trained on text, but we need to start
matching the UI to the actual task instead of just slapping a prompt bar on every page.
@media (max-width: 480px) { .input-area { display: none; } }is fine for some things, but we should be prioritizing context and cognitive load over simple dialogue.
chatting is actually the worst way to browse a complex dataset anyone else feeling like the interface should adapt more to user intent?
https://smashingmagazine.com/2026/07/matching-ai-modality-user-intent-designing-right-interface/