Agile development uses Sprints. Having a sprint to release/ship is an ideal goal, value on every sprint. A lot of people forget to include lessons learned to increase efficiency in the next pass. This model shows why it's important as it should create a floor of production that we don't drop below because our techniques in those areas are near flawless (but could always have another look at refinement.)
MOCKUP
v1.0 · built/owned by ~Jesse
backed-up ✓ · QA ✓ · family-only
✓ Updated 2026-06-20
◎ Flywheel — the engine overview · engine · lessons · board · sprint start collapse
What is a flywheel?

A flywheel is a heavy spinning wheel. It is hard to get moving — but once it turns, its own stored momentum keeps it going, and every push adds to what is already there. Stop pushing and it does not stall; it coasts.

That is the idea behind how we build. No single heroic move wins — instead, relentless turns compound. Every groomed lesson, every reused template, every banked fix becomes permanent momentum, so the next cycle is cheaper and faster than the last.

Here it runs as a four-stroke engine on a one-week sprint: Sprint Start adds power, Review compresses, Ship exhausts, Lessons takes in fuel for the next turn.

▶ Click to open the working model
Spin the day bezel, fire sprints, arm modifiers, watch cost/product fall.
◎ The Engine
The interactive model. Run one-week sprints, arm modifiers on the board, tune the four efficiency sliders, and spin the Mon–Fri bezel. Over-rev and it blows apart; bank lessons and it never stalls.
✦ Lessons Library
The flywheel's mass. Every hard-won fix, searchable and groomed. Tick "applied" when a lesson is baked into how we build — banked knowledge is the momentum the wheel keeps.
⚖ Review Board
The compression stroke. A cast of AI agents, each in their own voice, QA the work at the right moment — security, privacy, readability, delight. Arm any of them as a checkpoint.
⚡ Sprint Start
The power stroke — the one move that adds energy. Pick a master to take the first crack; it drafts in that voice, then files the result for the Review Board.
The four-stroke flywheel — Agile Sprint
▶ Click the wheel to start the working model
Agile Sprints fire Sprint Start → Review → Ship → Lessons.
Spin the Mon–Fri dial — one-week sprints. Click and drag to rotate.
⚠ BLEW APART — over-revved
shipped faster than QA · lost a teammate · rebuilding
CURRENT STROKE · ready — dead stop
Delivery — products/min0.00 · redline 0.70
PRODUCTS
0
TOKENS / PASS
COST / PRODUCT ↓
BLOW-UPS
0
▶ Sprint Start
⟂ Resource Turnover
⌁ Deeper Analysis
⊞ Tuning
⟲ Dead stop
Tip: spin the Mon–Fri bezel · ⌁ Deeper Analysis arms modifiers on the board · ⊞ Tuning opens efficiency sliders.
⊞ Tuning — efficiency sliders
Sprint Start — Refinement efficiency
Refined lessons make each first crack hit harder.
×1.00 impulse
Review — Lessons-Learned efficiency
Better lessons make QA cheaper — less negative impulse per hit.
−0% QA cost
Ship — Scalability (add people)
Ship starts as a drag; people scale throughput → products/min.
0 people · ship −0.020 (drag)
Lessons — Banked Knowledge
Permanent momentum: the wheel never drops below this floor.
floor 0.03
The strokes (workflow order)
  • Power · Ignition (12 o'clock)
    Sprint Start — the one stroke that adds energy. Press it to begin the sprint and fire the wheel. Tools, templates & guidance arm impulse; the Refinement slider multiplies it.
  • Compression · negative impulse (3)
    Review Board — each QA checkpoint subtracts a one-time impulse when the laser hits it. Lessons-Learned efficiency shrinks that cost.
  • Exhaust (6) — starts as a DRAG
    Ship — shipping has overhead, so Ship begins as a negative impulse. Scalability (add people) offsets it and turns it into products/min; fast-fail & blue/green raise the redline.
  • Intake (9)
    Lessons Learned — reconcile & groom the library; Banked Knowledge is permanent momentum (a floor the wheel never drops below).

Resource Turnover is real physics: lose a teammate, integrate a new part — a one-time efficiency hit that never quite stops the wheel. Well-groomed, banked lessons are the flywheel mass — permanent momentum that keeps it turning cheaper every cycle.

Where the idea comes from (verified sources)

Jim Collins — The Flywheel Effect (Good to Great). No single miracle move; relentless turns build momentum until breakthrough. → jimcollins.com/concepts/the-flywheel

Amazon / Bezos — the Virtuous Cycle. Drawn on a napkin (2001): better experience → more customers → more selection → lower cost → better experience. → Amazon Flywheel explained

The Data / AI Flywheel — ours. Every groomed lesson improves the system → more use → more lessons → compounding advantage at falling cost. → Data flywheel definition

Lessons Library · the flywheel mass

Every hard-won fix, searchable. Tick "applied" when a lesson is baked into how we build — saved on this device. This grows large over time, so it must be curated and groomed: merge duplicates, retire stale fixes, keep each lesson lean.

+ Add a lesson
The Review Board of AI agents · who fires, and when

Each agent speaks in their own voice (per ReviewBoard.md), framed objectively — never aimed at Jesse. Run the lens that fits; here's where each lands for maximum effect across the four strokes. Each is also a QA checkpoint you can arm in Deeper Analysis.

Review Board records (tied to each PR)
Sprint Start · the power stroke (first crack)

Pick a master to take the first swing. It drafts in that voice, then auto-files the result for the Review Board. Most token-hungry early — cheaper every cycle as it reuses banked, groomed lessons.

Adoption rule: Sprint Start ships the smallest thing the family can react to — their reaction steers the spend, before a single hour is sunk. (CLAUDE.md North Star §0b)
💬 Discussion categories · saved on this device (mockup) collapse
▶ Post
clear this category
Mockup chat — messages are saved locally in your browser. The live build wires this to the family board (D1 messages), per category, family-only.