Weekly Review. The Hypothesis Loop

15 minutes that close last week's loop and place this week's bets.

FoundationFoundation~7 min|5 checkpoints

By the end of this module

✓Completed a full 5-step weekly review

✓Wrote outcome-oriented intentions, not task-shaped ones

✓Promoted at least one weekly learning into a LESSON_LEARNED vault asset

✓Reviewed 4+ past snapshots and named one repeating pattern

✓Held the review cadence for 4 consecutive weeks without skipping

Why Review Is the Load-Bearing Habit

Every other Foundry feature degrades without the weekly review. Captures pile up. Commitments slip silently. Decisions accumulate without verdicts. Intentions drift into vagueness. Foundry becomes a backlog instead of a system.

The review is the cadence that keeps the system honest. 15 minutes, once a week, sets up the next 7 days and closes the previous 7. Without it, you're not running an operating system, you're running a digital filing cabinet that periodically yells at you.

If you skip everything else in this academy, do not skip the review habit. It's the one piece that makes the rest matter.

Watch out:If your weekly review cadence breaks for more than 2 weeks in a row, the system silently rots. Reinstating the habit is harder than maintaining it, protect the time block.

The Five Steps in Detail

Step 1. Commitments (2 min). Open the prior-week commitments list. For each: done, dropped, or rescheduled. Reschedule means notify the recipient before changing the date, silent reschedules destroy trust over time.

Step 2. Reflect (3 min). Free-form prose. The two productive questions are what surprised me? and what did I avoid? Both expose patterns that pure win/loss accounting misses.

Step 3. Clear Queue (3 min). Open RAW ideas. Triage them inline if quick; bulk park or archive the rest. The goal is to clear stale captures, if STALE_CAPTURES is firing every week, the imbalance is structural and review alone won't fix it.

Step 4. Wins & Learnings (3 min). Log wins to anchor a felt sense of progress. Promote at least one learning to a LESSON_LEARNED asset, this is the single highest-payoff habit in Foundry. Over a year, that's 50 distinct learnings, durably searchable.

Step 5. Intentions (4 min). The week's bets. See next section.

Intentions Are Hypotheses, Not Tasks

The most common review failure is writing intentions like tasks. The system invites this, they look like checkboxes, they sit on Today, you can mark them done.

But an intention is a bet on what matters this week, not a unit of work to complete. The distinction matters because the post-week question is different:

A task asks: did I do this?

An intention asks: was this the right thing to bet on?

Compare:

Task-shaped (avoid): "Write the landing page copy"

Intention-shaped: "Get the landing page to a state I'd show 10 people"

The intention-shaped version is harder to write because it forces a success condition. That's the whole point. At the end of the week, when you check it (or don't), you've learned something about your judgment, not just your throughput.

Pro plan users: click AI Summary before setting intentions. The AI reads the week's data and suggests 3 hypotheses based on what it sees. Treat them as a second opinion, not a script, adjust before saving.

→

Tip:A simple test: if an intention could be assigned to someone else and they'd know whether they completed it, it's a task in disguise. Real intentions require *your* judgment to evaluate.

Review History as Pattern Detector

/reviews shows your last 26 snapshots. The history is where the real learning lives.

Scan once a quarter. Look for:

Intentions you keep setting but never check. Either you don't actually care about them or your time/energy estimates are systematically off. Either way: useful information.

Weeks with no review. Drift periods. Cross-reference with project journals, did anything visible go wrong in those weeks? Usually yes.

Repeated wins. When the same theme shows up in wins 3+ weeks, that's a strength worth doubling down on, promote it to a STRATEGIC_PRINCIPLE asset.

Repeated learnings. When the same lesson shows up multiple times, you haven't actually internalized it. Time to capture it as a non-negotiable rule, not just a learning.

The review history is also where AI Pattern Detection pays back the most. It scans 60+ days of snapshots in seconds and surfaces patterns you'd miss reading them by hand.

Common Mistakes

Skipping a week. One week is recoverable. Two weeks is the danger zone, by week three you've lost the cadence and you'll need to consciously reinstate it. Protect the time block like a customer call.

Writing 10+ intentions. More intentions ≠ more focus. The cap is 5 for a reason. If you have 10 things that "must" happen this week, you have zero priorities. Cut to three.

Skipping Step 4 (Wins & Learnings). This is where the asset library grows. Skipping it for "speed" saves 3 minutes and costs you ~50 LESSON_LEARNED assets per year. Don't.

Treating the AI Summary as the truth. It's a second opinion based on data, not on judgment. Read it, disagree where appropriate, write your own intentions. The thinking is yours, not the AI's.

Practice Checkpoints

Completed a full 5-step weekly review

Wrote outcome-oriented intentions, not task-shaped ones

Promoted at least one weekly learning into a LESSON_LEARNED vault asset

Reviewed 4+ past snapshots and named one repeating pattern

Held the review cadence for 4 consecutive weeks without skipping

Next in Foundations Path

Capacity Governance. Saying No on Purpose

Why capacity is a constraint you set on yourself and how to defend it.

Start module