AI chat UX: keeping the plan in view (Paper cut #02)

You ask an AI to review your setup. It comes back with five numbered issues. You start on #1, and a few messages later the list is a screen and a half above you.

Now you're scrolling up to reread #3, writing "for #3…" and hoping the model knows which list you mean. By the end, you're not sure #4 ever came up again.

That's Paper cut #02, the second post in my series on small, fixable friction in AI products. The principle: when a reply hands the user a set of things to work through, the interface should hold on to it as work, not as text.

The full write-up on my site covers:

why this stays broken, even though everyone has felt it

a proposed fix: lists that pin, track progress and close when you're done

three ways to build it, from a front-end-only version to lists as conversation state

a 6-question audit you can run on your own product