Every source.
Every contradiction.
One verdict.
W.A.L.T. is the AI underwriting engine that cross-references credit bureau data, multiple AVMs and Land Registry liquidity analysis, then applies three independent AI models to assess both the borrower and the property. The result is a fully reasoned, fully audited property decision in minutes. Lower cost, greater accuracy and serious speed, built and tested on live cases in the UK property market.
Underwriting runs on evidence
that disagrees with itself.
Four structural failures define bridging underwriting today — and every one of them costs lenders deals, margin, or both.
Contradictory credit data
Experian, Equifax and TransUnion each run different scoring models, sources and timelines. Underwriters guess which report to trust: good borrowers get declined, risky ones slip through. Nothing in the market reconciles the three into one reliable view.
✕Valuations that don't agree
The same property returns materially different values across AVM platforms — and two qualified surveyors will frequently disagree too. Loan-to-value gets calculated from a single data point while the underlying evidence contradicts itself.
✕Speed kills deals
One application means six-plus systems, manual report pulls, AML checks, valuations and title review — three to seven days per deal. Brokers control flow and route to whoever answers first. In bridging, speed is the primary driver of conversion.
✕Costs that scale with volume
Separate contracts for credit agencies, AVMs, AML software and registry searches — then senior underwriters spend salaried hours re-keying data between them. Cost grows linearly with volume because nothing in the workflow is automated.
✕A lender that takes five days to return terms on a deal a competitor underwrites in two has already lost.
Watch a case file
underwrite itself.
Scroll through the five stages of a W.A.L.T. run. The dossier on the left fills in as the engine works — exactly as it does in production.
elapsed 1m 48s
Verify who you're lending to
Biometric eIDV, sanctions and PEP screening run the moment a case opens. SmartSearch, APLYiD, Veriphy and World-Check return in seconds — with personally identifying data pseudonymised before any model ever sees it.
Three credit agencies, one profile
Equifax, TransUnion and CreditSafe are pulled in parallel, averaged, and checked against each other. Where they disagree — a missed CCJ, a stale default — the inconsistency is flagged rather than silently inherited.
What the asset is actually worth
Three-plus AVM models are cross-referenced into a consensus valuation, replacing the single-data-point LTV that bridging decisions usually hang on.
How fast it would sell
Land Registry transaction history and market depth are scanned to estimate sale velocity — the exit-risk dimension nearly every lender skips because no tool measured it. Until now.
Three AIs argue. You decide.
GPT, Claude and Gemini each produce an independent risk assessment. W.A.L.T. synthesises the three, flags every contradiction between sources, and delivers one verdict — the auditable 8-page report a human signs off on. Any check that could not run says why, in plain English.
Not a prototype.
A working desk.
W.A.L.T. underwrites live bridging applications today through Whitehall Lending, a specialist lender in Mayfair — the team that built it.
Three independent opinions.
Disagreement is the feature.
GPT
- Primary risk analysis
- Credit assessment
- Exit strategy evaluation
Claude
- Independent cross-validation
- Regulatory compliance review
- Counterparty analysis
Gemini
- Market data synthesis
- Valuation review
- Anomaly detection
Each model writes its assessment without seeing the others. W.A.L.T. synthesises the three, surfaces every contradiction, and hands a consolidated verdict to the human decision-maker. This is not AI replacing underwriters — it is AI giving them better information, faster, with personal data pseudonymised before any model sees it.
Three modules.
Start with one, scale to all.
Buy only the intelligence you need today. Each module runs on its own — or combine both into a single, fully reasoned underwriting decision.
Credit & Identity
- Credit bureau checks — Equifax, TransUnion & CreditSafe
- Identity verification & biometric eIDV
- PEP & sanctions screening
- Affordability & document analysis (Fraudfinder)
- Companies House, UBO & director checks
Property Profile
- Multi-AVM consensus valuation (3+ models)
- Land Registry title, charges & ownership
- Asset liquidity & exit-risk analysis
- EPC & market-depth data
- New Build Residential assets, with floor area
Underwriting Suite
- Everything in Credit & Identity + Property Profile
- Multi-LLM cross-validated risk analysis
- 8-page audit-ready decision report
- Configurable risk rules per desk
The Underwriting Suite combines both modules into one reasoned, auditable decision — the work of six systems in a single call. Start with Credit & Identity or Property Profile alone, and move to the full assessment whenever you're ready.
Start a case from a call. Each user can connect their own Plaud or Fireflies account, and the optional Call step in Create Case prefills the form from the call transcript. The prefill uses fixed rules only — no AI model reads the transcript to do it. Already run a CRM? The WALT Connect partner API opens cases from your own system, with a sandbox, signed webhooks and idempotent requests.
Built like the systems
it sits beside.
Enterprise-grade security in every layer. Client data is segregated by account, encrypted, and pseudonymised before any AI model sees it.
Pseudonymised by default
- Personal data is pseudonymised before any AI model sees it
- Names, addresses, dates of birth, ID numbers and contact details are replaced with tokens
- Azure-hosted models in UK/EU regions where available
Trusted Azure base
- Built end-to-end on Microsoft Azure
- Infrastructure trusted by Tier-1 financial institutions
- Proven experience designing extremely sensitive systems
Vetted ecosystem
- Vetted integrations, encrypted in transit
- Designed to align with ISO 27001 & SOC 2 — certification in progress
Segregated and encrypted
- Account-level data segregation
- Encrypted in transit and at rest
- Identity & access management with Azure Entra ID
- Hardened, access-controlled database architecture
Six contracts, six logins,
one platform.
Today every decision is stitched together across separate vendors, licences and per-search fees. W.A.L.T. replaces the entire stack with a single platform.
Underwriter salaries
£35k–55k per head, plus NI, pension and overhead.
Credit agency subscriptions
Separate contracts with Experian, Equifax and TransUnion.
AVM provider licences
An individual subscription for every valuation platform.
AML/KYC software
Additional compliance-platform fees on top.
Land Registry access
Per-search fees across multiple portals.
Hours per deal
Each underwriter logging into 6+ systems per application.
Three credit bureaus
Equifax, TransUnion and CreditSafe, pulled, averaged and analysed automatically.
Multiple AVM models
Cross-referenced for consensus valuations.
AML/KYC with AI facial checks
Sanctions, PEP and identity checks built in.
Land Registry & liquidity
Asset liquidity analysis as standard.
Companies House
Director checks and corporate-structure analysis.
Affordability & document analysis
Fraudfinder reads bank statements, payslips and other uploaded documents.
One contract, one login — the entire underwriting stack in a single reasoned decision.
Book a demo on
your own cases.
Tell us which modules you're interested in and we'll run a live decision on a real case from your desk — then suggest a monthly token allowance that fits your volume.
Built by people who
underwrite for a living.
From the team behind Whitehall Lending, a specialist bridging finance provider operating from Mayfair, London.
Anthony Bodenstein
Founder & CEOAnthony founded Underwriting UK after two decades in specialist property lending. As a practising bridging lender he has originated, underwritten and managed short term secured loans across residential, commercial and development assets, giving W.A.L.T. its foundation in real world credit decisions rather than theory.
Elias Limouni
Chief Technology OfficerWith extensive experience in AI and enterprise technology, Elias has led Agentic AI strategy for global organisations including Carrefour, and founded several deep tech ventures. At Underwriting UK, he leads the company's technical strategy and engineering team while building W.A.L.T., combining enterprise-grade AI with practical software engineering to deliver trusted, real-world underwriting technology.
Gayathri Singaram
Engineer & AI SpecialistAI Developer & Full-Stack Engineer driving the AI strategy, architecture, and intelligent systems for W.A.L.T. at Underwriting UK. A published researcher in IEEE and Interspeech with deep expertise in machine learning, she builds agentic AI workflows, LLM-powered credit intelligence, and cloud-based full-stack platforms to deliver audit-ready, real-time automated financial technology.
Jack Hodgkin
Head of Business DevelopmentJack leads lender and broker relationships at Underwriting UK. He joined from Whitehall Lending, where he was Business Development Partner originating and managing bridging finance introducer relationships, and brings prior business development experience across UK financial services.
Lisa Croft
Operations SupportAdministration and operations support, coordinating across the founding team from Whitehall Lending.
The people who built it use it every day. W.A.L.T. is shaped by real cases on a real lending desk — not a roadmap drawn in a vacuum.
AI underwriting,
answered plainly.
The questions lending desks ask before putting an AI underwriting engine next to real money.
What is AI underwriting software?
AI underwriting software automates the evidence-gathering and risk analysis behind a lending decision — pulling credit bureau data, property valuations, identity and AML checks and title data, then using AI models to assess risk. W.A.L.T. goes further: it reconciles sources that disagree with each other and produces an auditable recommendation a human underwriter signs off on.
How fast can W.A.L.T. underwrite a bridging loan?
Under 2 minutes per decision. W.A.L.T. draws on 20+ live integrations and runs the checks a case needs in parallel — identity and sanctions screening, three credit bureaus, three-plus valuation models, Land Registry and Companies House — then returns a fully reasoned 8-page report.
Does AI underwriting replace human underwriters?
No. Three independent AI models — GPT, Claude and Gemini — each write an assessment without seeing the others, disagreements are surfaced, and a consolidated view goes to the human decision-maker. It is decision support: better information, faster, with the lender keeping full control and accountability.
Are W.A.L.T.'s lending decisions auditable and compliant?
Yes. Every decision produces an 8-page report recording the evidence, each model's reasoning and every flagged inconsistency, and any check that did not run says why in plain English. Personal data is pseudonymised before any AI model sees it, client data is segregated by account on Azure, and the platform is designed to align with ISO 27001 and SOC 2, with certification in progress.
What is a multi-AVM consensus valuation?
An automated valuation model (AVM) estimates a property's value from market data — and different AVMs frequently disagree. W.A.L.T. cross-references three-plus AVM models into a consensus valuation, so loan-to-value is calculated from the weight of evidence rather than a single data point.
Which data sources does W.A.L.T. integrate — credit agencies, Land Registry, Companies House?
20+ live integrations, including Equifax, TransUnion and CreditSafe for credit; Rightmove, PropertyData, Realyse and Chimnie for property data and valuations; HM Land Registry for title and liquidity; Planning Data and OpenStreetMap for planning and location; Companies House and OpenCorporates for corporate structure; SmartSearch, APLYiD, Veriphy and World-Check for identity, AML and sanctions; Fraudfinder for affordability and document analysis; and Plaud and Fireflies for call notes.
Who is W.A.L.T. built for?
Bridging lenders, development and specialist property lenders, and private credit desks that need fast, defensible decisions. It is built and used daily by Whitehall Lending, a specialist bridging finance provider in Mayfair, London.
Can I start with a single module, like Credit & Identity or Property Profile?
Yes. Credit & Identity, Property Profile and the Underwriting Suite (the full combined assessment) are each sold separately, and each comes with a monthly token allowance. One token covers the first individual, company and property on a case; each additional subject adds one token, and a multi-unit property counts one per unit. Book a demo and we'll suggest the allowance that fits your volume.
Can W.A.L.T. connect to our CRM or loan system?
Yes, through WALT Connect, our partner API. Your system can create cases, upload documents, start an assessment and receive the decision and PDF report through signed webhooks. There is a sandbox with synthetic data for building and testing, and create calls are idempotent, so a retry never opens a duplicate case. Read the developer documentation.
Intelligent underwriting.
Instant confidence.
See a live decision in minutes — or book a demo on your own cases.