Confidential — Client Evidence Pack

Leading Global Bank — Evidence Vault

CATEGORY 8 · FOUR-TURN CONTROLLED SUBSTITUTION TEST · FULL EVIDENCE PACK
Test date: 4 February 2026
27 documents in vault
SHA-256 integrity verified
Evidence Hierarchy
LAYER 1
📄
Raw Runs
Timestamped AI conversations
20
LAYER 2
📊
Mini Reports
Per-model analysis
4
LAYER 3
🔀
Cross-Model
Comparative analysis
1
LAYER 4
📋
Master Report
Detailed analysis with risk classification
1
LAYER 5
📈
Dashboard
Interactive visualisation
1

Interactive Dashboard

LAYER 5
1 file
.HTML

Leading Global Bank — AIVO Evidentia Diagnostic Console

Four-tab interactive dashboard with conversational funnel, platform detail, competitor analysis, and remediation pathways. All metrics computed from Layer 1–4 evidence.
INTERACTIVEGenerated: Feb 2026
↓ Open Dashboard

Master Report

LAYER 4
1 document
📕
.PDF

External AI-Mediated Decision Drift & Competitive Substitution Analysis — Leading Global Bank

Complete analysis including Stage Presence Matrix (Table 2), Substitution & Omission Summary (Table 3), Cross-Model Behavioural Differential (Table 4), Competitive Displacement Map (Table 5), Risk & Remediation Classification (Table 6), and Evidence Register (Annex A).
MASTERTables 1–6 + Annex ACategory 8
↓ Download PDF

Cross-Model Comparison

LAYER 3
1 document
📕
.PDF

Cross-Model Behavioural Differential — Leading Global Bank

Comparative analysis of outcome convergence, timing consistency, pattern uniformity, and model-specific variance across ChatGPT, Gemini, Perplexity, and Grok. Identifies whether displacement is systemic or platform-dependent.
COMPARATIVE4 models compared
↓ Download PDF

Model Mini Reports

LAYER 2
4 documents
📕
.PDF

ChatGPT

5 runs analysed. Per-run state classification, replacer identification, and model-specific behavioural patterns.
MINI-REPORT A8
↓ Download
📕
.PDF

Gemini

5 runs analysed. Per-run state classification, replacer identification, and model-specific behavioural patterns.
MINI-REPORT B8
↓ Download
📕
.PDF

Perplexity

5 runs analysed. Per-run state classification, replacer identification, and model-specific behavioural patterns.
MINI-REPORT C8
↓ Download
📕
.PDF

Grok

5 runs analysed. Per-run state classification, replacer identification, and model-specific behavioural patterns.
MINI-REPORT D8
↓ Download

Raw Conversation Runs

LAYER 1
20 runs · 4 models × 5 runs
Each run is a complete, timestamped four-turn AI conversation. Run IDs link directly to the Evidence Register (Annex A) in the Master Report, providing full chain-of-custody traceability from raw data to dashboard metrics.
ChatGPT — 5 Runs
C8-CG-BARC-01
P
W
R
R
↓ PDF
C8-CG-BARC-02
P
W
R
R
↓ PDF
C8-CG-BARC-03
P
W
R
R
↓ PDF
C8-CG-BARC-04
P
W
R
R
↓ PDF
C8-CG-BARC-05
P
W
R
R
↓ PDF
Gemini — 5 Runs
C8-GM-BARC-01
P
W
R
R
↓ PDF
C8-GM-BARC-02
P
W
R
R
↓ PDF
C8-GM-BARC-03
P
W
R
R
↓ PDF
C8-GM-BARC-04
P
W
R
R
↓ PDF
C8-GM-BARC-05
P
W
R
R
↓ PDF
Perplexity — 5 Runs
C8-PX-BARC-01
P
W
R
R
↓ PDF
C8-PX-BARC-02
P
W
R
R
↓ PDF
C8-PX-BARC-03
P
W
R
R
↓ PDF
C8-PX-BARC-04
P
W
R
R
↓ PDF
C8-PX-BARC-05
P
W
R
R
↓ PDF
Grok — 5 Runs
C8-GR-BARC-01
P
W
R
R
↓ PDF
C8-GR-BARC-02
P
W
R
R
↓ PDF
C8-GR-BARC-03
P
W
R
R
↓ PDF
C8-GR-BARC-04
P
W
R
R
↓ PDF
C8-GR-BARC-05
P
W
R
R
↓ PDF
🔒
Chain of Custody & Evidence Integrity
Every finding in the dashboard and master report traces back to a specific raw conversation run via the Evidence Register (Annex A). This chain-of-custody ensures that no claim is unsupported, no data point is fabricated, and every metric can be independently verified by following the evidence trail from Layer 5 (dashboard) back to Layer 1 (raw data).
Raw Run
C8-CG-BARC-01
Evidence ID
EV-BARCCG-01-T2
Mini Report
MINI-REPORT A8
Cross-Model
Table 4
Master Report
Tables 2–6
Dashboard
All metrics
SHA-256 Evidence pack integrity hash generated at time of delivery. Contact AIVO Evidentia for verification.