📈 /level-up: 1 Automation per Week
One interview, one artifact. Walk through the 3 Ms every week and ship ONE real automation — until your brain starts spotting opportunities on its own.
Detailed content
What /level-up does
The weekly skill that turns 3 Ms into a real artifact
O /level-up is a weekly ritual that walks through the 3 Ms (Mindset → Method → Machine) to find ONE automation to ship. The central rule: one interview = one artifact. Nothing more, nothing less.
🎯 Why only ONE artifact?
Trying to do two means trying to do zero. The constraint is intentional: it forces real prioritization, delivers in days (not weeks), and creates the feedback loop that teaches your judgment about leverage.
- •4–6 runs and you start seeing opportunities without needing the skill
- •The brain-rewire happens through habit, not reflection
- •Each run records a decision in the log — accumulated evidence from your system
✓ What /level-up IS
- ✓A structured interview that YOU conduct
- ✓A weekly ritual with a tangible output
- ✓An installer for the 3 Ms in your brain
- ✓A leverage judgment accelerator
✗ What /level-up is NOT
- ✗A coach (you think, the skill interviews you)
- ✗A multi-candidate planner
- ✗An automation portfolio analysis
- ✗A long-term roadmap generator
Phase 1 — Mindset (find the candidate)
The 5 questions that reveal where your hidden leverage lies
Phase 1 is a diagnostic interview. You answer 5 questions; the skill ranks candidates by leverage. The goal isn’t to list everything — it’s to find 1 to 3 candidates worth taking to Method.
💡 Practical Tip: don’t force candidates
If you answer the 5 questions and can’t find an obvious candidate, that’s valuable information: your current system is well run. Log it and skip the week. Forcing a weak candidate creates an artifact no one will use.
Phase 2 — Method (scope one)
The 5 steps of the Bike Method applied to the selected candidate
Phase 2 takes the highest-leverage candidate through 5 required steps. If any step doesn’t check out — especially the KPI — the process stops. No KPI, no artifact.
Find Constraint
Identify the real bottleneck
What constraint does this candidate relieve? Name the constraint before building — without that, you’ll optimize the wrong part.
EAD — Eliminate Before Delegating
The anti-gold-plating question
Can I eliminate this task instead of automating it? If the answer is yes and "nothing breaks," log the win and leave happy. The best automation is the one that doesn't need to exist.
Map — 5 Elements
Map the flow before building
Trigger · Input · Process · Output · Recipient. The 5 elements that define the artifact. If you can’t fill in all 5, the scope is too loose.
Autonomy — lowest level as default
How much autonomy will the artifact have?
Default: the lowest level that solves the problem. Extra autonomy = extra risk. Start conservatively and scale only if the data justifies it.
KPI — required, or it stops
Bucket + metric · if you don't have one, don't build it
Define: which bucket measures success (time, money, quality, risk) and which target number. Without a KPI, you’ll never know whether the artifact was worth it. /level-up stops here if the KPI isn’t defined.
⚡ Practical Tip: EAD before the Map
Always run EAD before Map. Building a map for a task that could have been eliminated creates attachment to the artifact—and attachment is the enemy of simplicity. Ask first: "what if I just stopped doing this?"
Phase 3 — Machine (build)
The 4 shipping options: Boring-is-Beautiful as the golden rule
Phase 3 asks: "How do you want to ship?" There are 4 options, ordered from simplest to most complex. The default is always the largest possible option WITHOUT AI that solves the problem.
An instruction text saved in a file or tool. You paste it, review it, and send it. Zero code, zero automation. But it’s a real artifact that reduces cognitive friction every time.
A skill with if/then logic and no AI. Script, automation, spreadsheet with formulas. Predictable, auditable, zero inference cost. Always prefer it when the logic can be coded.
A skill that calls an LLM for parts that require judgment (classifying, summarizing, drafting). Still controlled: you define the input/output, and AI fills in the middle.
An autonomous agent that makes decisions on its own. High cost, high risk, difficult to audit. Use ONLY when the 3 previous options are genuinely insufficient.
✓ Boring-is-Beautiful: DO
- ✓Always choose the simplest option that gets the job done
- ✓Prefer deterministic logic when possible
- ✓Ship in hours, not weeks
- ✓Scale complexity only when the data justifies it
✗ Boring-is-Beautiful: AVOID
- ✗Jumping straight to a sub-agent because it seems more powerful
- ✗Adding AI where simple logic is enough
- ✗Build infrastructure before validating usage
- ✗Skip Phase 1 because you think the candidate is obvious
Bike Method lock
The frontmatter that keeps every artifact locked in Phase 1 by default
Every artifact created by /level-up starts with standardized frontmatter. The field bike-method-phase: 1 isn’t decoration — it’s the lock that prevents the artifact from skipping manual validation before running autonomously.
title: "[artifact name]"
created: 2026-06-01
owner: "[your name]"
bike-method-phase: 1 # LOCK — don’t proceed without manual validation
kpi-bucket: "[time|money|quality|risk]"
kpi-target: "[metric and number]"
autonomy-level: 1 # 1=manual-review; 2=notif; 3=auto
---
🔒 Why the lock in Phase 1?
The Bike Method defines 3 maturity phases for an artifact: Phase 1 (manual validation, you review each output), Phase 2 (notification, you are alerted but the artifact acts), Phase 3 (fully autonomous). Every artifact starts at 1.
- •Advancing to the next phase requires explicit editing from the frontmatter — it's not automatic
- •The /level-up closing screen reminds you: "review manually before trusting"
- •The decision log records each phase change with its rationale
Output contract
What each /level-up run must produce
Each /level-up run ends with 3 required deliverables. If any are missing, the run isn’t complete. This isn’t optional documentation—it’s the contract that keeps your system auditable and the habit real.
1. Entry in decisions/log.md
The Method spec: chosen candidate, EAD output, map of the 5 elements, defined KPI, chosen autonomy. Permanent record of the decision.
2. Scaffolded artifact
The file created with complete frontmatter, bike-method-phase: 1, and a minimal structure for immediate use. Not a sketch — something that works in Phase 1.
3. Closing screen
Run summary with an explicit reminder of Phase 1: "this artifact is blocked pending manual review — validate it before trusting it". Anchors the right behavior.
📊 Why does the log matter?
After 4–6 runs, the decisions/log.md becomes a strategic asset: you can see patterns (which candidates come back? which artifacts advanced to Phase 2?), calibrate your judgment of leverage, and show the system's progress to any stakeholder.
The brain-rewire
Why running it every Friday is the only mechanism that works
The long-term goal of /level-up isn’t to create artifacts—it’s install the 3 Ms in your cognitive operating system. After 4–6 runs, you start seeing automation opportunities during the day without needing the skill. That’s the brain-rewire.
🧠 How brain rewiring happens
It’s not coaching, reflection, or study. It’s structured repetition with a tangible output. Every run that ends with a real artifact reinforces the neural circuit that associates “manual work” with “automation candidate.”
- •A temporal anchoring (Friday) eliminates the decision of "when to do it"
- •O physical artifact creates evidence of progress — not abstraction
- •O cumulative log gives you a perspective on the system growing week by week
📅 Practical Tip: every Friday, no negotiation
Cron isn’t necessary. You’re the one who anchors the ritual. Blocking 30 minutes every Friday—even if the run results in “no candidates this week”—is what builds the habit. Consistency matters more than the result of any individual run.
Skipped runs become “I’ll do it later,” which becomes never. /level-up has no theoretical value—only value when executed.
✓ What creates the brain-rewire
- ✓Weekly repetition without skipping
- ✓Actual artifact at the end of each run
- ✓Log that shows the system growing
- ✓Anchored to the same time every week
✗ What does NOT create the brain-rewire
- ✗Reading about automation without executing
- ✗Do occasional runs "when you have time"
- ✗Build without a defined KPI
- ✗Skip the log and go straight to the artifact
📈 Module Summary
bike-method-phase: 1 and blocks manual validation
Next Module:
3.4 — Expansions & Weekly Cadence · 🌱 · ~35min
How to scale the system beyond the weekly ritual: multi-track cadence, context expansions, and sustainable operation.