How I Use AI to Run My Weekly Review

How I Use AI to Run My Weekly Review

I always knew the weekly review was a good idea. I just never did it. It sat on my list of virtuous habits next to flossing and stretching — agreed-upon in theory, skipped in practice.

The reason was honest, if unflattering: my weekly reviews were boring and took forever. An hour of slogging through notes, half-remembering what happened, trying to extract lessons from a fog. Of course I skipped it.

Then I started running the review with an AI assistant, and the thing collapsed from a dreaded hour to a genuinely useful ten minutes. Now I actually do it. And the next week is measurably better for it, every time. Here's the exact process.

Quick Answer

My AI-assisted weekly review takes about ten minutes. I dump the week's raw material — calendar, notes, wins, frustrations — to an AI assistant, then ask it three things: what patterns it sees, what I should stop doing, and what the next week's top three priorities should be. The AI does the synthesis I used to slog through by hand. I make the decisions. Reflection becomes fast enough that I actually keep the habit.

Why most weekly reviews die

The weekly review fails for the same reason most good habits fail: the friction is higher than the felt reward in the moment.

You're tired, it's Friday, and the review asks you to manually reconstruct five days from scattered notes, then somehow extract wisdom from the pile. That's work — the slow, synthesizing kind — at exactly the time your brain wants to coast. So you skip it "just this week," and just-this-week becomes never.

The insight that fixed it for me: the valuable part of a review is the deciding — what to stop, what to prioritize. The tedious part is the gathering and pattern-spotting. And the tedious part is exactly what AI is good at. Offload the slog, keep the judgment, and suddenly the habit is light enough to survive. That same division of labor is what I keep coming back to in my honest take on which AI productivity tools actually earn their keep — let the machine handle synthesis, keep the decisions human.

Photo by Alpha En on Pexels

Step one: gather the raw material (2 minutes)

Garbage in, garbage out — so the review starts with a quick dump of the week's actual evidence. I don't curate. I just collect.

• My calendar for the week, copied in. This is the factual record of where time went.
• My notes — any #review tagged jottings I made during the week (an automation collects these for me, so this is instant).
• A two-line brain dump: what felt good, what felt frustrating. Raw and unfiltered.

That's the input. Two minutes, no thinking required — just gathering. The thinking comes next, and most of it isn't mine to do.

Step two: the three questions I ask the AI

This is the heart of it. I paste the raw material to an AI assistant and ask three specific questions, in order. The specificity is what makes it useful — vague prompts get vague reflections.

1. "What patterns do you see across this week?" The AI spots things I'm too close to see — that every frustrating day had back-to-back meetings, that my best work clustered on the two mornings I had no calls. Pattern-spotting across a messy week is exactly where a machine beats a tired human.

2. "Based on this, what's one thing I should stop doing next week?" Forcing a stop matters. Most reviews only add. This question makes the AI surface the lowest-value recurring drain so I can cut it.

3. "What should my top three priorities be next week, and why?" It proposes; I dispose. I rarely accept all three unedited, but having a concrete starting draft beats staring at a blank week.

I treat the answers as a sharp first draft from a thoughtful colleague, not as orders. The AI synthesizes. I decide. That division of labor is the whole trick.

The AI tells me what happened. I decide what it means.

Photo by Huy Phan on Pexels

Step three: turn answers into next week (3 minutes)

Insight that doesn't change next week is just journaling. So the last step converts the AI's draft into a concrete plan.

I take the agreed top three priorities and immediately block them into next week's calendar — named, defended focus blocks, the same way I learned to protect my best hours as real appointments, before the meetings arrive to claim the space. The one "stop doing" item becomes a literal rule for the week, written where I'll see it. Research summarized by the Harvard Business Review keeps landing on the same point: structured reflection is what turns raw experience into learning, and without it the weeks just blur.

This is where AI assistants and a little automation chain together nicely:

• The assistant drafts the calendar blocks from the three priorities; I approve them into place.
• A small automation files the week's review summary so I can compare across weeks and watch patterns build over a month, not just a day.
• Recurring "stop doing" items that show up three weeks running get flagged — a sign something structural needs fixing, not just willpower.

The whole thing, end to end, is about ten minutes. Two to gather, five to question and reflect, three to plan.

What changed once I actually kept it

The single biggest effect wasn't any one insight. It was continuity. Because the review was finally light enough to keep, the insights started to compound.

I could see a frustration recur across three weeks and finally fix its root cause instead of re-discovering it each Friday and forgetting by Monday. The AI's pattern-spotting across months of stored reviews surfaced things no single week could — a slow drift, a creeping commitment, a recurring energy dip tied to a specific meeting.

A reflection you do once is a nice afternoon. A reflection you do every week, lightly, for a year, quietly reshapes how you work. The AI didn't make me wiser. It made the wise habit cheap enough to actually have.

The prompts that actually work (and the ones that don't)

Since people always ask for the exact wording, here's what I've learned about prompting an AI assistant for reflection — because the quality of the answer lives or dies on the quality of the question.

The biggest lesson: vague prompts get vague reflections. "How was my week?" produces a bland, useless summary, the productivity equivalent of "fine, thanks." The fix is to give the AI a role, real evidence, and a specific job. Instead of asking it to opine, I ask it to analyze concrete material I've handed it and answer a pointed question.

A few patterns that consistently produce sharp answers:

• Anchor it to evidence. "Based on this calendar and these notes, what pattern do you see?" beats "what should I focus on?" The first forces grounded analysis; the second invites generic advice.
• Force a single answer. "What's the one thing I should stop?" outperforms "what could I improve?" Constraint produces clarity. An open question produces a list you'll ignore.
• Ask for the why. "Suggest three priorities and explain why each matters next week" makes the reasoning visible, so I can disagree intelligently instead of accepting or rejecting a bare list.
• Make it compare. Once you have a few weeks stored, "what's recurring across these three reviews?" surfaces drift that no single week reveals. This is where the AI genuinely outperforms a human — holding many weeks in view at once.

And the prompts that don't work, that I've abandoned:

• Anything that asks the AI to make the decision for me. It can't know my values; when I let it, the plan felt hollow and I didn't follow it.
• Anything generic enough to apply to anyone. If the answer would fit a stranger, the prompt wasn't specific enough about my actual week.
• Long, rambling prompts with ten questions at once. The AI scatters. One sharp question at a time, in sequence, keeps the answers crisp.

The meta-lesson is that the AI is a mirror, not an oracle. Feed it a blurry, generic question and it reflects blur back. Feed it your real, specific, slightly embarrassing week and a pointed ask, and it reflects something you can actually use.

If a ten-minute reflection sounds worth trying, run the three questions this Friday and stick around for more field notes on building light, durable habits that actually survive a busy week.

The bottom line

The weekly review never failed because it was a bad idea. It failed because it was too heavy to keep, and a habit you can't keep is worth nothing.

Make the good habit cheap, and you'll finally do it. AI made my review cheap enough to survive contact with a tired Friday, and a year of small reflections beats a dozen ambitious ones I never did.

What would change if you actually reviewed your week — every week, in ten honest minutes? Try it once this Friday. Just three questions. See what it tells you.

Key Takeaways

• Offload the tedious parts of your weekly review—gathering raw material and spotting patterns—to an AI assistant, keeping only the decision-making for yourself to reduce the process to ~10 minutes
• Start with a 2-minute raw material dump: copy your calendar, any tagged notes, and a two-line brain dump of wins and frustrations—no curation, just collection
• Ask the AI three pointed questions in sequence: (1) 'What patterns do you see across this week?', (2) 'What’s one thing I should stop doing next week?', (3) 'What should my top three priorities be next week, and why?'
• Turn the AI’s answers into action by immediately blocking the top three priorities into your calendar and writing the 'stop doing' item as a visible rule for the week
• Use specific, evidence-anchored prompts (e.g., 'Based on this calendar and notes, what pattern do you see?') to avoid generic advice—vague questions yield vague reflections
• Store weekly reviews to let patterns compound over time; recurring 'stop doing' items after 3+ weeks signal structural issues worth fixing, not just willpower problems