Weekly Review is the ritual that makes GTD work — and the one almost everyone skips. Mind Like Water reads your week and writes you a clear summary. You still make every decision. The app just does the reading for you.
Try it free
David Allen calls weekly review the "critical success factor" for GTD. Read his book and the message is clear: this is where the system actually works or doesn't. Then close the book and ask the next ten people if they actually do one. They'll laugh.
It's not that people don't want to. It's that doing one cold takes an hour, costs serious mental energy, and starts with staring at a long flat list of stuff. By the time you've decided what to look at first, you're already drained.
AI Weekly Review changes the starting point. Instead of a blank page, you start with a draft.
At the top of your Weekly Review, click Generate with AI. After a few seconds, four things appear:
What's not sent: your email address or identity, completed tasks older than the review window, notes outside the review window, anything from other users.
First time you click Generate, you'll see a consent modal explaining what's sent. Accepting writes a flag to your profile. You can revoke anytime in Settings → AI Review. Anthropic doesn't retain the request content.
Every "AI:" pill is editable. Every commitment-extraction is shown with the quoted source so you can verify before accepting. The narrative summary is a draft you read, not an answer you accept on faith. The decisions are yours; the reading is the AI's.
The summary names specific projects ("the kitchen remodel moved forward, the Q2 launch plan is stuck") because the model is given specific project data. It doesn't say "you had a productive week" — it says what actually happened, in your own taxonomy.
The Notes → Tasks extraction is the unique value. Most review tools surface what's in your task list. This one reads what you wrote in your daily pages and finds the things that should be tasks but aren't yet. It's the safety net for the "I meant to follow up on that" gap.
The limits are mostly there to keep the costs reasonable. For weekly use, you'll never hit them.
Four things: a 3-paragraph narrative summary of your week (what you shipped, what drifted, one suggested focus for next week); per-item suggestions on overdue, stale, and someday tasks (each shows as an "AI:" pill you tap to accept); commitments pulled from your recent daily pages that you probably should turn into tasks; and one-sentence observations when an Area went quiet or showed unusual activity.
No — it accelerates it. The discipline is still yours: looking at where you are, deciding what's next, deciding what to drop. The AI just reads the data and drafts the framing so you start with a half-built review instead of a blank page. You make every actual decision.
Your active tasks (title, status, project, area, dates), your active projects (name, outcome, task count, recent activity), recent notes from the last 7-14 days (title + body preview), and your Areas of Focus with activity counts. Completed tasks go in too so the "what you shipped" section can be written. Anthropic doesn't retain the request content.
31 generations per month per Pro user, 5 per hour. Re-running within 15 minutes returns the existing review; after that, a fresh one is generated.
14 days free, no credit card. Run a weekly review at the end of the week. See how much faster it goes when you're not starting from scratch.
Start free trial Free for 14 days. Pro is $8/month after.