Methodology · v0.1

The AJIR rubric

Your AIQ is multiplicative across four pillars. Weakness in one cannot be hidden by strength in another. The score is anchored to evidence — a self-attested win counts less than an audited one.

AIQ = 200 + 800 × (A0.30 × J0.25 × I0.30 × R0.15)

Pillars

A

Applied AI Capability

Weight 0.3
  • Depth
    Real solved problems, not prompt dumps.
    35%
  • Variety
    Distinct artifact types — repeating one kind is penalised.
    25%
  • Workflow integration
    AI embedded in actual work, not isolated.
    25%
  • Continuity
    Sustained submissions over time.
    15%
J

Human Judgment

Weight 0.25
  • Validation
    Did you check what the model produced?
    30%
  • Refinement
    What you changed, and why.
    25%
  • Decision rationale
    Why this output, this audience, this moment.
    25%
  • Override quality
    Where you disagreed with the model.
    20%
I

Verified Impact

Weight 0.3
  • Magnitude
    Time saved, users helped, decisions improved.
    70%
  • Evidence confidence
    Multiplier from L1 self-attest → L5 audited.
    30%
R

Responsible Deployment

Weight 0.15
  • Governance
    Policy, scope, sign-off.
    35%
  • Privacy
    PII / data handling.
    25%
  • Explainability
    Can you defend this to the affected party?
    20%
  • Human review
    Required for high-risk outputs.
    20%

Lenses

Lenses translate the AJIR pillars into the languages you build in.

Craft blend

A 35 · J 40 · I 10 · R 15

AJIR weighting for craft & taste work

Outcome blend

A 15 · J 35 · I 40 · R 10

AJIR weighting for decisions & framing

Builder blend

A 55 · J 20 · I 15 · R 10

AJIR weighting for tool depth & systems

Voice blend

A 30 · J 40 · I 20 · R 10

AJIR weighting for story & attention

Evidence ladder

Impact magnitude is multiplied by the strongest evidence you can show.

L1Self-attested×0.4
L2Screenshot / file×0.6
L3URL crawl (live page)×0.7
L4Org-confirmed×0.85
L5Independently audited×1

Tiers

Apprentice200–399
Practitioner400–599
Operator600–749
Architect750–899
Pioneer900–1000

Anti-gaming

  • • Repeated artifact types in a row reduce variety credit.
  • • Unsupported impact claims are scaled down by the evidence multiplier.
  • • Prompt dumps without judgment reflection cannot exceed Apprentice.
  • • High-risk artifacts without governance evidence cap R at 0.4.
← See your AIQYour certifications →