P
Current organisation
Pilot Client
pAIsley Governance Platform

See the risk. Test the governance. Strengthen what matters.

A working pilot environment for pAIsley’s Ethical AI Health Check and related advisory offerings. It records evidence, applies transparent decision rules and keeps professional judgement visible.

Health Check
0%
survey + lens map completed
Maturity position
Not assessed
5 & 3 Lenses · Health Check v2.0
Priority findings
0
items requiring attention
Red maturity cells
—
5 × 3 cells requiring most work
VALUE SCORECARD
AI MATURITY REVIEW
Five leadership questions · aggregated across Bias, Explainability and Privacy & Data Ethics
5 × 3 ETHICAL AI MATURITY MAP
FIVE STAGES × THREE LENSES

Ethical AI Health Check

Live

Establish the maturity baseline, identify where attention is needed and produce a client-ready dashboard.

Governance Stress-Test

Live

Apply governance to a real AI use case: harms, safeguards, human oversight, escalation and response.

Frontier Model Fitness

Prototype

Capture independent evidence and assess which model is fit for this organisation, use case and risk.

Trust & Informed Consent

Prototype

Test transparency, meaningful choice, contestability and stakeholder expectations for an AI use.

Ethical AI Sounding Board

Live

Create a structured decision record for opportunities, concerns and governance questions needing independent challenge.

Te Ao Māori Fitness

Next

Designed as a distinct fitness lens for tikanga, relational fit, te reo, rangatiratanga/data sovereignty and governance fit — not a token composite score.

pAIsley journey
HEALTH CHECK
STRESS-TEST
WEAVE (OPTIONAL)
BLUEPRINT
OPERATIONALISE
MONITOR
This pilot supports pAIsley judgement; it does not provide certification or automated assurance.
Waypoint #1 · Prepare

Ethical AI Health Check

The canonical pAIsley Health Check v2.0 prepared for the encasedIT portal. It uses the five leadership questions across the three lenses of Bias, Explainability and Privacy & Data Ethics, with responses grounded in current practice rather than future intention.

Assessment profile

0 of 37 questions assessed21 maturity-scored + 16 qualitative questions
Governance in practice

AI Governance Stress-Test

Select a live or emerging AI opportunity and test whether the organisation can explain the value, see the harms, apply safeguards and respond when a boundary is crossed.

Use case

Preparedness Network

Who needs to be able to act when a difficult issue arises?

Governance test

pAIsley Frontier Model Rating · PFMR

Model Fitness Assessment

Not “which model is smartest?” — which model should this organisation trust for this use case? The assessment separates independent evidence, client-specific model fitness and Te Ao Māori fitness so technical capability cannot override material governance or cultural concerns.

Advanced: use after changing the selected model’s detailed inputs.
Start here

Describe how the organisation intends to use AI

This is the main workflow. pAIsley will infer the use-case profile, set the initial weighting, assess and rank the current models, run the provisional Te Ao Māori comparison where relevant, and identify only the questions that still need confirmation.

Advanced / audit view — evidence, deployment, weights and governance gates
What this section is for: the main analysis above compares all candidate models automatically. Use this advanced view only after selecting one model if you want to verify or change its evidence, deployment assumptions, PFMR weights, governance gates or cultural inputs. Then use Recalculate selected model to update that one model’s detailed recommendation.
1
Independent evidenceWhat credible external evidence says about this model/version.
2
PFMRHow fit the model is for this client, use case and risk profile.
3
Te Ao Māori FitnessA distinct cultural, sovereignty and governance assessment where relevant.

Assessment context

Deployment profile

How will the client actually use the model?

0 / 6 known

These details can materially change Privacy, Security and Operational Fit even when the underlying model is identical.

The evidence library pre-populates a provisional public-evidence grade. It sits beside the PFMR rather than being hidden inside it, and can be adjusted if client-specific or newer evidence materially changes the picture.
Layer 1

Independent evidence register

0 / 7 sources checked

Select a model from the evidence library to load pAIsley's current public evidence baseline. Exact-version coverage varies by source; gaps and proxy evidence are shown explicitly.

Public evidence baseline last checked: 23 August 2026. “Source checked” does not mean that the source has evaluated the exact current model. Where only an earlier model or developer-level measure is available, the note identifies it as proxy evidence or a coverage gap.
Layer 2

PFMR evidence-led dimensions

Weights 100%

Scores are 0–100. Weighting can be adjusted for the client/use case; the calculation normalises the entered weights.

Governance gates

Capability cannot override these

An unresolved material gate can override the numerical PFMR recommendation.

Layer 3

Te Ao Māori Fitness Assessment

A distinct assessment rather than a single “Māori score”. The simple workflow now pre-populates a provisional assessment from the intended use; this advanced view records the five dimensions, evidence and gates that require validation.

Fitness dimensions

Cultural / sovereignty gates

For a Relevant or Material assessment, unresolved gates are surfaced explicitly and can make the overall recommendation conditional or require escalation.

This assessment supports informed judgement. It is not a proxy for Māori participation, appropriate cultural expertise, consultation, consent or endorsement where those are required.
PFMR result
Complete the evidence inputs, weighting and governance gates.
Te Ao Māori position
Assess applicability and the cultural fitness dimensions.
Overall recommendation
Calculate the full assessment to combine the technical, governance, evidence and cultural layers.

Assessment logic

Independent evidence

Artificial Analysis, Stanford HELM, METR, MLCommons, Stanford FMTI, UK AI Security Institute and NIST AI RMF form the evidence stack used by the pAIsley framework.

pAIsley judgement

Evidence grade, client-specific weighting and governance gates turn a generic benchmark into a defensible use-case recommendation.

Cultural fitness

Te Ao Māori considerations remain visible as a distinct judgement layer. A high PFMR does not cancel a material rangatiratanga, data sovereignty, tikanga or governance concern.

Independent challenge

Ethical AI Sounding Board

Capture the issue, competing considerations, decision and follow-up so advice becomes part of the governance record rather than disappearing into email or meetings.

Issue intake

Decision records

No records saved yet.
pAIsley administration

Platform configuration

Change selected Health Check wording, dashboard labels and maturity thresholds without editing the source code.

Prototype note: this local HTML file has no real authentication. In the hosted portal, this page should be restricted to pAIsley Admin users and methodology changes should be logged.
Dashboard settings

Stage labels

These five labels appear in the Value Scorecard and the 5 × 3 maturity map.

Scoring settings

Maturity colour thresholds

0 is weakest and 4 is strongest. Set the upper boundary for each band before green.

Health Check v2.0

Question & lens editor

Edit all 37 Health Check questions and their Bias, Explainability and Privacy & Data Ethics prompts.

Client-facing wording

Dashboard titles

Methodology control

Current baseline

Source methodologyHealth Check v2.0 · encasedIT
ScaleN/A + 0–4
Leadership stages5
Ethical lenses3
Questions37

Kerry can export a configuration file, retain versions, and re-import it later without changing the app code.

Client output

Health Check Report

Generated from the pAIsley Health Check v2.0 responses. Maturity scoring is transparent; interpretation and recommendations remain subject to pAIsley professional judgement.