The app is open and signed in. It doesn’t know anything about you yet, and that’s the next thing to fix.
A conversation with Claude Code is a back-and-forth in the chat pane. You ask; Claude does whatever it needs to do to answer — usually reading something — and tells you what it found.
Point it at a folder
There’s an Open folder button on the welcome screen. Pick a folder you don’t mind Claude poking around in: your Downloads, a folder of notes, a side project.
Pick one you already know. Being able to check the answer is what makes this first conversation worth anything.
The folder’s files appear down the side of the window, and the chat fills the middle. That folder is now Claude’s whole world.
Ask it something
Click into the chat box and type:
what's in this folder?Press enter, and watch — don’t skip this bit. You’ll see roughly:
- A short thinking indicator.
- A line saying it’s reading the folder.
- A summary in plain English: “This folder has three subfolders and a README…”
That sequence is the whole rhythm of using Claude Code. You ask, it thinks, it goes and looks, it answers. Everything else on this path is a variation on it.
Ask a follow-up
Don’t start over. Ask something that builds on what just happened:
what does the README say?This time Claude won’t re-read the whole folder — it already knows the layout from ten seconds ago. It goes straight to the file. The conversation has memory, the way a conversation with a person does.
Try a few of your own. Notice that the answers are about your files, with your file names in them, not a generic answer that would fit any folder. That’s the tell that it actually looked.
When to start fresh
New topic, different folder, or the conversation has wandered somewhere unhelpful? Start a new chat. The old one is set aside and the new one begins with no memory of it.
That’s a feature. A fresh chat is a Claude that isn’t carrying assumptions from the last thing you did.