The weekly review with Claude: a system for actually thinking about your week

Most people who try to do a weekly review give up on it within a month. The ritual sounds valuable in theory — reflect on the week, plan the next one, stay aligned with what matters — but in practice it feels like a lot of thinking and writing for a document nobody reads, including yourself.

The problem is not the weekly review concept. It is the format. Writing a weekly review for an audience of one, where the "output" is a journal entry that lives in a folder, is not compelling enough to make the habit stick.

Using Claude changes the format. Instead of writing to yourself, you are having a conversation that produces specific outputs — a clear-eyed account of what happened, a prioritized list for next week, and decisions that actually get made rather than deferred again. Twenty minutes, every Friday. Here is how.

The setup

Use a dedicated conversation in a Claude Project for your weekly reviews. This is important: having the context of previous reviews available means Claude can notice patterns across weeks that you would not catch on your own. By month three, you have a Claude that knows your recurring bottlenecks, your energy patterns, and which commitments you consistently overestimate.

Name the project something you will actually open: "Weekly Operating System" or just "Friday Review."

The conversation structure

Part 1: The honest account (5 minutes)

Start by telling Claude what actually happened this week — not what you planned, what happened. This is a data dump, not a polished narrative. Be specific and include the things that did not go well.

Prompt: "Here is my week. Tell me: what patterns do you see? What did I spend time on that was not in my plan? What did I avoid that probably matters?"

Claude is good at noticing the gap between what you said you would do and what you actually did. That gap is almost always the most important thing to examine.

Part 2: The processing pass (5 minutes)

Now clear the decks. What is hanging over from this week that needs to be resolved before next week starts?

Prompt: "Here are the things still in my head that are not done: [list]. Help me decide for each one — do I do it now, schedule it, delegate it, or drop it?"

This is the GTD capture step, done with a thinking partner rather than alone. You will make decisions faster because Claude will push you to be concrete: "Do it now" means you do it today. "Schedule it" means you name the day. "Delegate" means you name who. "Drop it" means you say out loud why it is not worth doing.

Part 3: Next week (5 minutes)

What matters most next week? Not everything you have to do — the three or four things that, if they get done, will make the week a success regardless of what else happens.

Prompt: "Based on what I've told you about my priorities and what slipped this week, what should the top three focuses be next week? What should I protect time for? What should I push back on or say no to?"

Claude will sometimes push back on your priorities. When it does — when it says "you've mentioned X three weeks in a row as a priority and it keeps not happening, is it actually a priority or is it something you feel guilty about not doing?" — that is the value of having context across weeks.

Part 4: The one decision (5 minutes)

Every good weekly review surfaces one thing you have been avoiding deciding. Not an action item — a decision. Should you end this client relationship? Should you change how you are approaching this project? Should you have a conversation you have been putting off?

Prompt: "What is the one thing I'm clearly avoiding deciding, based on what I've told you this week and in previous weeks? What would it look like to make that decision today?"

You do not have to decide it in the review. But naming it out loud, with Claude reflecting it back to you, removes the ability to pretend it is not there.

What you leave with

At the end of twenty minutes you should have:

• A clear account of what actually happened this week (not a sanitized version)
• A resolved list of loose ends — each one either scheduled, delegated, or dropped
• Three specific priorities for next week
• One decision named, even if not yet made

This is not comprehensive. It is not a productivity system. It is the minimum viable weekly review — the version that is honest enough to be useful and short enough to actually do.

The longer-term payoff

After eight to ten weeks of using the same Project, Claude starts to see things you cannot see in a single session. It will notice that you consistently overcommit on Mondays, that your energy for strategic work disappears in the afternoon, that a specific recurring meeting is consistently listed as a frustration but never examined.

These patterns are hard to see week-to-week. They become obvious in the aggregate. That is the payoff of keeping the reviews in the same Project instead of starting fresh each time.

Further reading

• What's new in Claude: Turning Claude into your thinking partner — how Claude can serve as a reflection partner