At a glance
What this is: Two settings most Claude users have not found. A monthly recap of how you actually use it, and optional reminders to stop.
Who it is for: Anyone who has wondered whether they lean on Claude more than they meant to.
Where to look: Settings > Reflect for the recap, Settings > Time and focus for the reminders.
Most apps are not built to be used less. Claude now ships two settings that point the other way: a monthly report on how much you use it and what for, and a separate set of controls that nudge you to stop.
Most people have not gone looking, because both live in settings and are still rolling out. They are among several summer changes that got an announcement post and not much else, which we track in what is new in Claude in 2026.
What does the recap actually show you?
Go to Settings > Reflect on the web or the desktop app. It loads the past month by default, and a control at the top switches the range to this month so far, or the past three, six or twelve months.
- An opening summary naming the actual work and themes from your conversations.
- Three numbers: your most active day, your peak hour, and total conversations, over a chart showing how they spread across the period.
- What you spent time on: a proportional breakdown of your themes, each with a description and a percentage.
- Expanding your skills: observations pulled from your own chats, with a concrete next step attached to each.
Your phone activity counts toward all of it, even though the page itself is not available on mobile.
What four skills does it look for?
That last section is where the feature shows its hand. Anthropic organises its observations around four skills:
- Delegation, meaning what to hand off and what to keep.
- Description, meaning the context you give Claude.
- Discernment, meaning weighing what comes back.
- Diligence, meaning how you represent AI use in your work.
Two of those four are about weighing the output rather than taking it, and about being straight that you used AI at all. Most vendor marketing does not put doubt and disclosure at the centre of what competence means. It matches what we have argued for a while: the hard part is judgement, not phrasing. Our own take is in the AI literacy skills worth building.
Break reminders and quiet hours
These live separately, under Settings > Time and focus, and work whether or not you ever open the recap.
A break reminder nudges you once you have spent a set amount of time with Claude in a day. You pick the threshold yourself, and when you reach it Claude shows a prompt you can snooze or dismiss. Quiet hours work by day of the week with a start and end time. Open Claude during one and it reminds you that you set the time aside, and you can carry on if you meant to.
Neither is a lock. Anthropic is explicit that you can always continue past both. You set them on web or desktop and they apply across web, desktop and mobile.
On demand is not the same as not collected. Anthropic says the report is only generated when you open Settings > Reflect. That is worth knowing, but it does not mean nothing is being recorded: the recap is built from the chat history that memory already keeps, and that accumulates whether you look or not. What happens on demand is the summary, not the collection.
What does it leave out, and why does that matter?
Left out of the recap entirely: chats you started in incognito mode, any conversation that pulled in health data from Apple Health or Health Connect, and everything you do in Claude’s two separate products, Cowork and the developer tool Claude Code.
Connected services get a specific rule. If Claude reads your Gmail or Google Drive, the raw emails and files do not enter your recap, but anything Claude wrote in the chat about them can. Ask it to summarise your inbox and the summary counts as chat content, even though the emails do not.
Anthropic says sensitive and distress-related topics are treated differently: they are not used to lead the recap and are not broken out with counts or percentages, and support resources may appear alongside. That is a deliberate limit on a feature whose whole job is to show you your own habits.
| Plan or detail | Monthly recap | Break reminders and quiet hours |
|---|---|---|
| Free, Pro, Max | Yes, in beta | Yes, in beta |
| Team, Enterprise | Not available | Not documented |
| Where you set it | Web and desktop | Web and desktop |
| Where it applies | Counts mobile activity, page not on mobile | Web, desktop and mobile |
| Needs memory on | Yes | No |
Why can you not see it?
Four things to check:
- It has not reached you. It is in beta and rolling out gradually.
- Memory is off. The recap is built from the same chat history that powers memory, so with memory off the Reflect page does not appear. There is no separate switch. See what Claude remembers about you.
- You are on mobile. Web and desktop only.
- You are on a work plan. The recap is not available on Team or Enterprise at all. Only Free, Pro and Max.
The exclusion worth noticing is the work plans. Those are the accounts where people tend to use Claude most hours per week.
What a recap cannot tell you
- It counts conversations. It has no idea which ones earned their place. Twenty short chats that saved you an afternoon look the same as twenty you could have skipped.
- It cannot see the work you did without it, which is the number that would actually tell you something.
- A nudge you dismiss every day is not a boundary. Turning it on takes ten seconds. Honouring it does not.
The settings nobody tells you about
One issue every morning, plain English, for people who use AI at work.
Free forever. Unsubscribe anytime.
Common questions about the Claude recap
Does looking at my recap use up my usage limit?
No. Viewing it costs nothing against your plan. If you start a conversation from one of its suggestions, that chat counts like any other.
How do I turn the recap off?
There is no dedicated switch. It is tied to memory, so turning off Generate memory from chat history in Settings also switches off the recap. Anthropic does not say what happens to recaps you have already viewed. That trades a lot to get a little, so consider whether simply not opening the page is enough.
Do break reminders stop me using Claude?
No. Both are prompts you can walk past. You can snooze, dismiss or continue every time.
Is this the same as usage limits?
No, and Anthropic separates them explicitly. Usage limits are what your plan allows. These are boundaries you set for yourself, and they have no effect on your allowance.
Can my employer see my recap?
No. The recap is not available on Team or Enterprise plans, so there is nothing for a workspace admin to look at. If you use a personal Free, Pro or Max account for work, the recap belongs to that account, not to your employer.
Sources
- Anthropic — See your monthly recap (9 July 2026)
- Anthropic — Set break reminders and quiet hours (9 July 2026)
- Anthropic — A new way to reflect on how you use Claude
All settings and availability above were checked against Anthropic’s help centre on 31 August 2026. The feature is in beta and rolling out, so what you see may differ.
Read next
- What Claude remembers about you
- The AI literacy skills worth building
- How to turn AI off, everywhere
- What is new in Claude in 2026
- How to use Claude AI, from scratch
- What Claude Cowork actually does
- The AI glossary
Two ways to go further
The AI Prompt Library
1,000+ ready-to-use prompts for Claude, ChatGPT, and Gemini. Stop staring at a blank box.
Get it for $39 →2-Hour Live AI Crash Course
A private, beginner-friendly session across Claude, ChatGPT, Gemini, and the wider landscape.
Book for $125 →