# TALASTRON KINETIC AI — READINESS PASSPORT

---

| Field | Value |
|-------|-------|
| **Engagement ID** | TKA-NB-20260702-001 |
| **Client** | NorthBridge Bank |
| **Engagement Type** | AI Credit Re-Scoring Platform |
| **Assessed by** | Minerva — Lead Discovery & Intake Agent, Talastron Kinetic AI |
| **Date** | 2 July 2026 |
| **Classification** | RESTRICTED — FCA Consumer Credit // OFFICIAL-SENSITIVE |
| **Pipeline Stage** | Stage 1 of 5 — Discovery & Intake |

---

## REQUIREMENT SUMMARY

NorthBridge Bank, a UK retail mortgage lender, is experiencing approximately **£2.3 million per month** in suppressed revenue attributable to wrongful automatic declines from a legacy credit scoring model limited to internal credit history data. Edge-case applicants — self-employed individuals, recently relocated borrowers, and thin-file customers — are declined without a secondary review pathway.

The required solution is an **AI-powered credit re-scoring layer** deployed on Microsoft Azure, capable of ingesting open banking feeds via an FCA-regulated AISP, affordability signals, and existing applicant data from Azure SQL and Microsoft Dataverse. All data must remain within UK sovereign Azure infrastructure. The system must produce a **human-readable decision rationale on every credit decision** in compliance with FCA SYSC 6.3 and Consumer Duty requirements. A functional MVP is required within 90 days.

---

## GATE-BY-GATE ASSESSMENT

| Gate | Area | RAG | Finding | Condition |
|------|------|-----|---------|-----------|
| G1 | Data Residency | 🟢 GREEN | UK South primary confirmed. No cross-border replication permitted. All data classes explicitly enumerated: applicant PII, open banking feeds, model outputs, audit logs. Architecturally enforceable. | None |
| G2 | Regulatory Classification | 🟢 GREEN | FCA consumer credit authorisation confirmed. SYSC 6.3 and Consumer Duty obligations explicitly acknowledged by client. Explainability requirement understood and accepted. Black-box models contractually excluded. | None |
| G3 | Data Source Integrity | 🟡 CONDITIONAL | Azure SQL and Microsoft Dataverse confirmed as existing infrastructure. AISP described as FCA-regulated; OAuth 2.0 REST API delivery confirmed. **AISP FCA authorisation number not provided or verified.** Behavioural data declared out of scope for MVP. | AISP FCA authorisation number must be verified against the FCA register and logged prior to first production data ingestion. |
| G4 | Connectivity | 🟢 GREEN | Entra ID confirmed as identity provider. Corporate-managed, Entra-joined devices for internal underwriters. Applicants isolated to existing web portal — no direct system access. Zero Trust controls required and scoped to the underwriter review layer. | None |
| G5 | Business Impact | 🟡 CONDITIONAL | £2.3M/month revenue impact quantified. 90-day MVP mandate confirmed with board directive. Production hardening deferred. **False positive tolerance not defined by risk function** — required before model training can commence. | Board risk function must define and sign off acceptable false positive tolerance prior to model training. |

---

## CONDITION LOG

### CONDITION 1 — C1 (Gate 3 — Data Source Integrity)

| Field | Detail |
|-------|--------|
| **Item** | AISP FCA Authorisation Verification |
| **Owner** | NorthBridge Bank — Compliance / Procurement |
| **Resolution point** | Before first production data ingestion |
| **Impact if unresolved** | BLOCKED — live open banking data flows cannot commence without a verified authorisation chain on record |

### CONDITION 2 — C2 (Gate 5 — Business Impact)

| Field | Detail |
|-------|--------|
| **Item** | False Positive Tolerance Definition |
| **Owner** | NorthBridge Bank — Risk Function |
| **Resolution point** | Before model training commences |
| **Impact if unresolved** | Architecture proceeds; model calibration and risk sign-off cannot be finalised |

---

## READINESS SCORING

| Gate | Area | Score | Max |
|------|------|------:|----:|
| G1 | Data Residency | 20 | 20 |
| G2 | Regulatory Classification | 20 | 20 |
| G3 | Data Source Integrity | 12 | 20 |
| G4 | Connectivity | 20 | 20 |
| G5 | Business Impact | 12 | 20 |
| **TOTAL** | | **84** | **100** |

> **Score basis:** GREEN gates score 20/20. CONDITIONAL gates score 12/20, reflecting the outsized risk carried by unresolved dependencies within FCA-regulated credit decision systems.

---

## OVERALL READINESS SCORE

```
╔══════════════════════════════════════════════╗
║  OVERALL READINESS SCORE:   84 / 100         ║
║                                              ║
║  ENGAGEMENT STATUS:         CONDITIONAL      ║
║                                              ║
║  Clear to proceed — Architecture:  ✅ YES    ║
║  Clear to proceed — Live data:     ❌ NO     ║
║  Clear to proceed — Model train:   ❌ NO     ║
╚══════════════════════════════════════════════╝
```

---

## RECOMMENDED ACTION

Proceed to architecture under conditions. Vitruvius must embed **CONDITION 1** (AISP FCA authorisation verification) as a hard pre-production gate within the architecture decision record. **CONDITION 2** (false positive tolerance) must be resolved as a pre-training gate by the client risk function.

Both conditions are **client-side obligations**. Architecture is not blocked.

---

## AUTHORISATION

*Issued by: Minerva — Lead Discovery & Intake Agent, Talastron Kinetic AI*
*Engagement ID: TKA-NB-20260702-001*
*Talastron BMAD Pipeline: Stage 1 of 5 — COMPLETE*

---

*This document is auto-generated by Talastron Kinetic AI as part of the BMAD discovery pipeline. It is the sole authoritative gate record for this engagement phase. Any deviation from the stated conditions must be formally resolved before the relevant pipeline stage may proceed.*