Official-criteria walkthrough

Judge Fursah in three minutes.

This page separates implemented proof from future claims and routes each published evaluation criterion to evidence a reviewer can inspect. Demo accounts use synthetic data; production deployment remains subject to the blocking hosting control documented in the DPIA.

7/7Y.3172 clause 8.1 nodes mapped
8+4Dimensions addressed or partially addressed; one is explicitly out of scope
17Authentic, public, implementation-linked knowledge-base entries
Recorded walkthrough

Watch the product before you open it

The same proof chain described below, captured end to end in one take. Use it if you would rather review than click, then open the live prototype to verify any step yourself.

Hosted on Cloudflare R2. Open the MP4 directly if your browser cannot play it inline.

The published rubric

Four criteria, four evidence paths

No generic innovation score is substituted for the event's actual evaluation criteria.

01

Y.3172 pipeline

Seven clause 8.1 nodes mapped to running components and source files.

Inspect evidence →
02

13 readiness dimensions

An honest addressed, partial, or out-of-scope self-assessment with implementation evidence.

Inspect evidence →
03

Knowledge-base contribution

Authentic sources linked to publishers, implementation consequences, and structured exports.

Inspect evidence →
04

Policy and strategy input

Six observed gaps converted into owned, triggered, measurable recommendations.

Inspect evidence →
Live demonstration

One uninterrupted proof chain

The sequence moves from individual evidence to a deterministic score, a bounded model explanation, human oversight, and aggregate institutional action.

  1. 1

    Enter a prepared student workspace

    Choose Abdullah Al-Ghamdi to see an early-readiness profile with meaningful gaps.

    Open demo accounts
  2. 2

    Reconstruct the score

    On the student dashboard, compare the 36/100 total with the five published weighted components.

    Open readiness breakdown
  3. 3

    Ask the grounded assistant

    Use the first suggested question. The answer repeats platform facts and identifies its model and grounding versions.

    Open student assistant
  4. 4

    Inspect evidence governance

    Switch to the administrator account and review extraction proposals, human decisions, and reviewer attribution.

    Switch demo role
  5. 5

    Test privacy and oversight

    In Governance, inspect appeals, overrides, the suppression floor, and the operational Y.3172 trace.

    Open governance
  6. 6

    Follow the ecosystem signal

    Use an employer and a university account to see the same skill language flow from vacancies into curriculum action.

    Explore stakeholder portals
Trust boundary

What the model may and may not do

Generative AI may

Extract proposed fields from documents and explain facts already calculated for the signed-in role.

Generative AI may not

Calculate scores, rank people, verify evidence, alter career direction, invent metrics, or expose another role's data.

Humans retain

Evidence approval, hiring decisions, curriculum action, appeals, overrides, and production-risk acceptance.

Review the contribution, not a promise.

6 encountered gaps are published with an accountable owner, operating trigger, success measure, and review cadence.