This week's focus: always tell the user what's happening.
Last week we talked pricing. This week it's the thing that quietly makes or breaks trust in any app, and it matters even more for the AI apps you're building: letting people know the app is working, and telling them clearly when something goes wrong.
Quick refresher from last week, how you show the price matters as much as the price itself:
This is the #1 usability rule from the Nielsen Group: visibility of system status. The app should always keep people informed about what's going on, with clear feedback, quickly.
It splits into two jobs, and both matter double for AI apps, because AI takes a few seconds to respond:
Show that the app is working while it thinks, so no one stares at a blank screen wondering if it froze.
When something breaks, say what happened in plain language and what to do next, never a scary code or a dead end.
Pivotal = build this in. Basics = the simple staples. Polish = nice later, skip for now.
The response appears word-by-word as it's generated, instead of making them wait for the whole thing.
The little dots or "thinking…" shown while the AI or an agent is working, so it doesn't feel frozen.
When something fails, show a plain message and a way to try again, instead of crashing or going blank.
The error shows right at the field as they type, not after they submit.
The little spinning icon that means "working on it."
Shows how far along a task is, best when you know roughly how long it'll take.
Gray placeholder shapes of the content while it loads. Feels faster than a spinner.
Show the result instantly (assume it worked), and quietly undo it if it fails. Feels instant.
A small message that pops up briefly to confirm an action or an error, then disappears.
Drop this into Claude Code and it looks at what you're building, then adds the right feedback for it, streaming, a thinking indicator, and clear errors, wherever a user could be left staring at a frozen screen.
Look at what I'm building and tell me where a user could be left staring at a frozen or blank screen. Then add proper feedback: stream the AI's responses so they appear as they're generated, show a "thinking" indicator while an agent is working, and give clear, plain-language error messages that say what happened and what to do next. Never let the user wonder if it's broken.
That's the move, the AI checks your specific build (chat, voice, or agent) and recommends exactly where it needs feedback.
As of August 2, the EU's AI transparency rules are enforceable, and they reach the US. The law follows the user, so if someone in the EU can reach your bot, you're in scope. Several US states and the FTC already require it too.
Disclosure is a trust feature that happens to be the law. And you don't have to check it by hand, let the AI do it:
Check this app for AI disclosure compliance and fix what's missing. Make sure: (1) every chat or voice interface says it's "AI" in its first message and shows a visible AI label on screen, not just in the terms; (2) any AI-generated image, audio, or video is clearly labeled; (3) any published AI text that informs people has either an "AI-generated" label or a named human reviewer. Tell me what's missing, add it, and flag anything that would conflict, then propose a compliant fix instead of dropping it.
Tip: paste this same instruction at the start of a new project too, and it builds disclosure in from day one instead of after.
Plain-language summary, not legal advice. Anyone with real EU revenue or enterprise clients should have counsel review their setup.
Post your wins, show what you're building, and let's make it feel alive together. This is what we do, week by week, inside the community.
Join us →