Trustworthy prioritisation by design

Turn Fleet ideas and backlog into trusted priorities

One living compass for Fleet Mortgages within Starling Group to capture colleague ideas and technology roadmap items, test simplest solution fit, score value, impact and return on investment and route safe 90-day shadow tests.

Illustrative pre-workshop calibration. Synthetic data only. Lending policy and credit decisions remain 100% human-led.

01 ONE FRONT DOOR

Unify executive examples, colleague suggestions and existing engineering or operations backlog items in one shared register.

02 VALUE AND ROI ENGINE

Weight six themes and eighteen level-down checks across culture, colleagues, brokers, workflow, 90-day wins and low risk.

03 GOVERNED ROUTING

Balance weighted value against post-control residual risk across four clear delivery tracks with named feedback.

Executive proposition

Unify ideas and roadmap items under one front door

Replace informal filtering and competing queues with one transparent operating rhythm ahead of the 10 December 2026 workshop.

End silent idea loss

Give colleagues and executives visible criteria and constructive feedback so good operational ideas do not stall expecting a no.

Avoid false tech trade-offs

Assess standard process fixes, clean data and API rules alongside AI assistance instead of forcing separate queues.

Score practical return on investment

Reward broad colleague reach, cleaner end-to-end case flow, low-cost 90-day wins and low residual risk without inventing cash savings from released minutes.

Protect human service pride

Keep underwriters, BDMs and operations leads in control of lending judgement and external broker communication.

Ready for 10 December 2026

Synthetic cohort baseline

Illustrative options
9
Intake themes
5
Hard gates
4
Weighted ROI themes
6
Delivery tracks
4

Qualification and scoring

Filter four hard gates before scoring six ROI themes

Every option clears four sequential blockers first, then earns a calibrated 1, 3 or 5 rating across eighteen level-down checks.

  1. Gate 1

    Strategic and problem fit

  2. Gate 2

    Simplest solution check

    Routes non-AI fixes directly to Process or API track

  3. Gate 3

    Permitted data and platform readiness

  4. Gate 4

    Named human accountability

1. Culture and mindset

15%
  • Builds openness and belief in change
  • Counters fear of a no or job loss
  • Preserves Fleet human service pride

2. Colleagues and reach

20%
  • Touches many roles and functions
  • Cuts repetitive work and grows skills
  • Frees net time after human checking

3. Customer and broker

20%
  • Makes broker submission steps easier
  • Improves landlord speed and certainty
  • Aids BDM reach and fair outcomes

4. Operations and workflow

20%
  • Optimises end-to-end case workflow
  • Cuts rework and missing-info loops
  • Reuses clean data across applications

5. Quick win and ROI

15%
  • Delivers a visible win within 90 days
  • Low cost using existing tools or data
  • Strong return vs build and run effort

6. Low risk and readiness

10%
  • Low residual risk inside appetite
  • Testable now on safe synthetic data
  • Protects core origination delivery

Themes 1 to 4 carry 75% stakeholder and operational impact. Themes 5 and 6 carry 25% practical ROI and readiness. A failed hard gate is never overridden by a high score.

Trust and residual risk

Cut residual risk across six UK lending pillars

Practical controls bring promising options inside Fleet and Starling risk appetite before any shadow test begins.

1. Safety and resilience

Offline shadow boundary, fallback queues and operational resilience aligned with FCA guidance.

2. Transparency

Source-grounded outputs citing exact Fleet policy clauses, dates and case documents.

3. Consumer Duty fairness

Consistent broker and landlord support aligned with FCA PRIN 2A fair outcomes.

4. SMF accountability

Named Senior Management Function sponsor and human sign-off aligned with FCA October 2026 AI governance findings.

5. Data governance

UK GDPR lawful basis, synthetic test packs and zero unpermissioned third-party harvesting.

6. Human contestability

Underwriters and BDMs review, edit or override every suggestion with one click.

Governed workspace

Run five portal views to route every option cleanly

Fast-track shadow test

90-day offline synthetic test.

Process or API track

CTO or Operations engineering backlog without AI overhead.

Targeted discovery

Time-boxed permission, consent or baseline scoping.

Park with revisit log

Halted at a hard gate or parked transparently with a clear revisit trigger.

Launch interactive Sentinel Nexus portal

Lending policy, underwriting judgement and external broker or borrower communications stay 100% under Fleet human sign-off.