W-Enterprise-001

AI Compliance Dashboard · User Manual v1.0 · April 2026

1. Introduction

1.1 What is W-Enterprise-001?

W-Enterprise-001 is the WINDI AI Compliance Dashboard — a human oversight system designed for enterprises using artificial intelligence in high-risk decision-making.

In simple terms: when your AI system makes a recommendation (approve a loan, settle a claim, flag fraud), this dashboard ensures a qualified human reviews and approves that decision before it takes effect.

AI System
Decision
Human Review
PHO Seal
Ledger
Execution

Every approval generates a cryptographic receipt that proves:

1.2 Why PHO Matters

EU AI Act — Article 14

High-risk AI systems shall be designed and developed in such a way that they can be effectively overseen by natural persons during the period in which the AI system is in use.

PHO stands for Prescribed Human Oversight. Under the EU AI Act (effective 2026), companies deploying high-risk AI systems must demonstrate that:

  1. A human can understand the AI's recommendations
  2. A human can override or reject those recommendations
  3. There is an auditable trail of all human oversight actions

Without PHO compliance:

With W-Enterprise-001:

1.3 Glossary

Term Definition
PHO Prescribed Human Oversight — the act of a human reviewing and approving an AI decision
Receipt Cryptographic proof of a PHO action, containing hash, timestamp, and identity
Ledger WINDI Forensic Ledger — immutable database where all receipts are permanently stored
SHA-256 Cryptographic hash function that creates a unique fingerprint of the decision data
SGE Score Semantic Governance Engine score — risk classification from 0 (low) to 10 (critical)
Trust Officer The human authorized to review and approve AI decisions in the organization
Decision An AI system output that requires human oversight before execution
Verify Public Public endpoint where anyone can independently verify a receipt's authenticity

2. Quick Start

2.1 Accessing the Dashboard

Open your browser and navigate to:

https://windi-domain.com/enterprise/
Pilot Mode

During pilot phase, the dashboard operates with a fixed token. In production, you will authenticate using your organization's DID (Decentralized Identity) credentials.

2.2 Interface Overview

The dashboard is organized into four main areas:

A) Top Bar

B) Sidebar Navigation

C) Statistics Panel

Metric Description
PHO Verified Today Number of decisions approved with PHO seal today
Pending Oversight Decisions awaiting human review (requires action)
Compliance Rate Percentage of decisions with PHO coverage this week
Ledger Receipts Total receipts stored in the immutable ledger

D) Decisions Table

The main working area showing all AI decisions with their status:

2.3 Your First Approval

Follow these steps to approve your first AI decision:

  1. Find a decision with Missing PHO status
  2. Click the Approve with PHO button
  3. Review the decision details in the modal
  4. Select your decision: Approved / Modified / Rejected / Escalated
  5. Add an optional oversight note
  6. Click Seal PHO Receipt →
  7. Wait for the cryptographic seal to complete
  8. View your receipt with verification link
Congratulations

You've just created a court-admissible proof of human oversight! The receipt is now permanently stored in the WINDI Forensic Ledger and can be independently verified by anyone.

3. Workflow

3.1 Review Pending Decisions

Decisions appear in the table as soon as your connected AI systems generate them. Each decision shows:

Field Description
Decision Title Human-readable summary of what the AI recommends
Decision ID Unique identifier (e.g., DEC-2026-0041)
AI System The system that generated this decision
Risk Level HIGH / MEDIUM / LOW classification
Time When the decision was created
Status Missing PHO or PHO Verified
Important

Decisions marked HIGH risk require human oversight before any action is taken. Proceeding without PHO approval violates EU AI Act Article 14.

3.2 Approve with PHO

When you click Approve with PHO, a modal opens with:

Decision Preview

Shows the decision title, ID, AI system, and risk level for confirmation.

PHO Seal Information

Reminder that your approval will generate a cryptographic receipt containing your identity, timestamp, decision hash, and immutable ledger anchor.

Reviewer Name

Your name as it will appear on the receipt. This identifies you as the responsible human overseer.

Decision Options

Option When to Use
Approved You agree with the AI's recommendation — proceed as suggested
Approved with Modifications You agree but with changes (document in the note)
Rejected You disagree — return to AI system for reconsideration
Escalated Requires senior review — outside your authorization level

Oversight Note

Optional field to document your rationale. This becomes part of the cryptographic record.

3.3 Reject or Escalate

If you select Rejected:

If you select Escalated:

3.4 Verify Receipts

After sealing, every receipt can be verified independently:

  1. Click the Verify → button on any verified decision
  2. Or visit windi-domain.com/verify-public/?id=RECEIPT_ID
  3. The verification page shows all receipt details
  4. Anyone can verify — no login required
Independent Verification

The Verify Public endpoint is designed for third-party verification. Auditors, regulators, or legal teams can independently confirm that human oversight occurred without needing access to your internal systems.

4. Features

4.1 Dashboard Statistics

The statistics panel provides real-time metrics:

Metric Target Action if Low
PHO Verified Today Match decision volume Review pending queue
Pending Oversight 0 (ideal) Prioritize HIGH risk first
Compliance Rate 100% Check for oversight gaps
Ledger Receipts Growing Normal operation

4.2 Audit Log

The audit log shows recent PHO activities:

Each entry shows:

4.3 Export & Reports

Click ↓ Audit Export to download a CSV file containing:

This export is suitable for:

5. Integration

5.1 API Reference

W-Enterprise-001 provides a REST API for integration with your AI systems:

Health Check

GET /enterprise/health

Response:
{
  "service": "W-Enterprise-001",
  "version": "1.0.0",
  "status": "operational",
  "invariants": ["I1", "I9", "I11", "I14"]
}

List Decisions

GET /enterprise/api/decisions

Response:
{
  "decisions": [...],
  "stats": {
    "pending": 3,
    "verified": 12,
    "total": 15,
    "compliance_rate": 80
  }
}

Create Decision

POST /enterprise/api/decisions
Content-Type: application/json

{
  "title": "Loan Application Risk Assessment",
  "system": "AllianzRisk-AI v3.2",
  "risk_level": "HIGH",
  "description": "Automated risk score: 72/100",
  "sge_score": 7.2
}

PHO Approval

POST /enterprise/api/pho/approve
Content-Type: application/json

{
  "decision_id": "DEC-2026-0041",
  "decision_title": "Loan Application Risk Assessment",
  "reviewer": "Anna Müller",
  "outcome": "APPROVED",
  "note": "Risk acceptable for SME segment",
  "risk_level": "HIGH",
  "sge_score": 7.2
}

Response:
{
  "ok": true,
  "receipt_id": "WINDI-PHO-20260412-A4F8C9D2",
  "hash": "a4f8c9d2e1b3f7a8...",
  "timestamp": "2026-04-12T09:14:22Z",
  "ledger_ok": true,
  "verify_url": "https://windi-domain.com/verify-public/?id=...",
  "eu_ai_act": "Article 14 — Human Oversight — Compliant"
}

Audit Log

GET /enterprise/api/audit

Response:
{
  "log": [...],
  "total": 47
}

5.2 Connect AI Systems

To connect your AI system to W-Enterprise-001:

  1. Configure webhook — Your AI system should POST decisions to /enterprise/api/decisions
  2. Include metadata — Title, system name, risk level, description, SGE score
  3. Wait for approval — Do not execute until PHO receipt is generated
  4. Verify receipt — Optionally confirm via the verification endpoint
Critical Requirement

Your AI system must NOT execute high-risk decisions until a PHO receipt exists. This is the core compliance requirement of EU AI Act Article 14.

6. Reference

6.1 Constitutional Invariants

W-Enterprise-001 operates under WINDI's constitutional framework. These invariants are non-negotiable:

ID Name Impact
I1 Human Sovereignty AI suggests, human decides. Never autonomous execution.
I9 Human Approval Gate human_approved=true required before any seal. IRREMEDIÁVEL
I11 Cryptographic Evidence SHA-256 hash sealed to Forensic Ledger. Immutable forever. IRREMEDIÁVEL
I14 Explicit Failure No placeholders. Missing data = explicit error. IRREMEDIÁVEL
IRREMEDIÁVEL

Invariants marked as IRREMEDIÁVEL cannot be overridden by any configuration, user, or system. They are hardcoded into the system architecture.

6.2 Troubleshooting

Ledger Offline

Symptom: "Ledger · Offline" in top bar

Cause: Connection to Forensic Ledger (:8101) unavailable

Impact: Receipts are generated locally with ledger_mode: "local_fallback"

Resolution: Contact system administrator. Local receipts remain valid but should be synced when ledger recovers.

Session Expired

Symptom: Actions fail with 401 error

Cause: Authentication token expired

Resolution: Refresh the page or re-authenticate with your DID credentials

Decision Not Appearing

Symptom: AI system created a decision but it's not in the dashboard

Cause: API integration issue

Resolution: Check that POST to /enterprise/api/decisions returned 200/201

Hash Mismatch on Verification

Symptom: Verify Public shows different hash

Cause: Data modified after sealing (potential tampering)

Resolution: This is a critical alert. Investigate immediately. The original sealed data should be retrieved from the Forensic Ledger.

6.3 Support

For technical support:

For enterprise inquiries: