Evidence & Checkpoints
Status: The gates, not the rituals.
Purpose
This is where the program stops trusting your attendance and starts trusting your work. Every section ends at a decision gate: you cannot usefully continue until you can produce the artifact with the verification the gate requires.
How a gate works
Each gate names three things:
- The artifact — what you produce.
- The verification — how you prove it's real work, not a confident paste.
- The consequence — what you are not ready for until the gate is passed.
You do not get a badge for passing a gate. You get the right to continue. That's the whole reward, and it's the one that matters.
The gate rules
- Gates are passed with artifacts and evidence, not with quiz scores or completion checkmarks.
- AI use is allowed — this is a program about AI use. Disclosure is required. A polished output without traceable reasoning is insufficient evidence.
- "I do not have enough evidence" is a complete and acceptable answer at any gate. Pretending otherwise is the real failure mode.
Section gates
| Section | Gate artifact |
|---|---|
| 01 Core Judgment | The ten-question analysis of a real AI decision you faced, with your answers, your evidence, and what would change your mind |
| 02 Personal Effectiveness | The personal method + the baseline you can defend — your module 5 one-page method, applied to a task you had never used AI for before |
| 03 Workflow & Systems | The redesigned workflow + its measurement — the smallest useful change, its before/after numbers, and the failure simulation you ran with the real people |
| 04 Building & Agents | The working build + its evaluation record — the solution contract, the narrow-agent design, and the verification record with its explicit ready / needs changes / not ready decision |
| 05 Leadership & Adoption | The decision brief with a named owner — the ten-question gate answered at org level, with the recommendation and the ownership map |
| 06 Enablement | The transfer artifact + its evidence — the capability statement, and the sample of real participant work scored against it |
The standard
Every artifact at every gate is evaluated with the Verification Rubrics — the Core Rubric for artifact quality and the Checkpoint Rubric for advancement. Passing the Core Rubric means at least "Acceptable" on every row and "Strong" on Accountability and Verification. The Checkpoint Rubric means the filters are applied unprompted, the artifacts meet the standard, the person can explain what would change their mind, and fluent AI output is never treated as finished work.
Running the program
- Facilitators and program owners: the one-pager on running the program without turning it back into theater — the success measure, the rubrics in practice, Core as a hard gate, evidence review without bureaucracy, and the phases.
- Governance: the practical reference for NIST AI RMF, ISO/IEC 42001, and the EU AI Act, plus the lightweight operating model (inventory, classify, own, contract, measure, review).
- Rollout: the implementation timeline — prepare, Core Judgment, first applied packs, governance, enablement, steady state.
- Tool notes: the overlays section — thin, temporary, replaceable. Tool-specific notes live there on a 90-day review cadence. The core never depends on them.
- Learners use the rubrics to self-check before calling anything done. Reviewers use the same rubrics so standards stay consistent. Program owners use the timeline and governance model to avoid both chaos and bureaucracy. Everyone treats external frameworks as tools, not as the goal: better decisions with clear ownership.