GUIDES
Claude Just Built AI Screen Time 📚
Claude just launched Reflect. It is like Spotify Wrapped for your AI habits, and it quietly shows you exactly where your work should be automated. Here is how.
Ten years ago, Apple gave us Screen Time. Not because anyone asked for it, but because nobody realized they were on their phone five hours a day. You cannot fix what you cannot see.
Claude just did the same thing for AI. It is called Reflect, and it is basically Spotify Wrapped for how you use Claude. It shows you your busiest day, your peak hour, your most common topics, and the tasks you keep repeating. Anthropic rolled it out in the past week, and it is the first time a big AI company has handed you a mirror like this.
Here is the part I care about, and it is why I wanted to write this. Reflect is not really about productivity, and it is definitely not about guilt. It is about awareness, and then automation. Once you can see the same task showing up over and over in your week, you are looking at a map of exactly what to hand off.
What Reflect Shows You
Open Reflect and you get a clean look-back on how you have actually been using Claude. Here is everything it pulls together for you.
- Your most active day. The single day you leaned on Claude the hardest.
- Your peak hour. The time of day you reach for it most, so you can see your real work rhythm.
- Total chats. How many conversations you have had, all in one honest number.
- Your most common topics. The subjects you keep coming back to.
- Tasks you repeat. This is the important one. The kinds of work you ask for on a loop, week after week.
You can set the window to the past 1, 3, 6, or 12 months, so you can zoom in on a busy week or look at a whole year at once. Anthropic has also said a view of how much time you have spent is coming next, which will make the Screen Time parallel even more exact.
The Nudges
Reflect does not stop at the numbers. It watches for patterns and gives you a small nudge when it spots one, and this is where it gets genuinely useful.
If you keep asking for the same kind of task, it suggests turning that into a Project, so you stop starting from scratch every single time. If you have been going for hours, it offers to turn on quiet hours or schedule break reminders, so the tool bends around your life instead of the other way around.
How To Turn It On
It takes about a minute, and you can do it on the web or in the desktop app.
- Open your Settings.
- Go to Reflect.
- Make sure Memory is turned on. Reflect reads from your memory, so if memory is off, it has nothing to look back on.
Right now Reflect is in beta, and it is open to Free, Pro, and Max users, so you can try it no matter which plan you are on.
How I Actually Use It
Here is how I actually use Reflect, and it is not to feel bad about a big chat count. I go straight to my most repeated tasks. That list is really a to-do list for automation. Anything I do over and over is something I should build once and then never think about again. Here are two of my own examples.
Say Reflect shows that most mornings you paste in a rough email and ask Claude to clean it up. That is a Project waiting to happen. Build one with your tone, your rules, and a few sample emails saved inside it. Now you drop in the messy draft and it comes back sounding like you, every time, with no re-explaining.
Say your repeated tasks show you summarizing the same kind of thing every week, like meeting notes or client calls. Turn it into a Project or a saved skill that runs the summary the exact same way each time. You paste the notes, you get the format you always want, and you skip the setup completely.
That is the whole shift. Screen Time made us aware of our phones. Reflect makes you aware of your AI, and then it hands you the exact list of things to automate. Read your top repeated tasks, pick one, and build it this week. That is how a look-back turns into real time back in your day.
The real unlock of AI agents is not full autonomy, but human-in-the-loop systems design. Structure your agent inputs with explicit constraints, negative prompts, and automated test checkpoints. When building tools, keep token consumption lean by caching system prompts and isolating tasks into specialized sub-agents.