Automated KYC/AML with a complete audit trail
When the FCA asks how a compliance decision was made, show them exactly how. Veritas runs each check through a documented rule engine and logs everything.
Compliance you can prove.
Book Your Shadow PilotFor CEOs & Founders
Compliance doesn't have to be a cost centre. Automate the manual bottleneck, onboard customers faster, and pay per check instead of per analyst.
Lower Customer Acquisition Cost
Manual compliance review is a hidden tax on every new customer. Automate the bottleneck and convert applicants faster at a fraction of the cost per check.
Faster Onboarding
85% manual → exception-only
Go from 85% manual review to exception-only handling. Applicants clear compliance in seconds, not days. Less drop-off, faster time-to-revenue.
Measurable ROI
283 analyst-hours/month saved at 1,000 users/month
At 1,000 applicants per month, that's 283 analyst-hours back. Time your team can spend on growth instead of copy-pasting names into screening tools.
Predictable Unit Economics
Per-check pricing that scales linearly. Your compliance cost shrinks as a percentage of revenue as you grow, instead of scaling with headcount.
For CROs & MLROs
Not a score and a confidence level. A documented decision with a full audit trail you can present to the FCA on demand.
FCA Audit Readiness
When the FCA asks how a compliance decision was made, you need more than "an analyst reviewed it." The system logs the input data, the rules applied, and the outcome for each check. That record is available instantly.
SM&CR Personal Liability
Under SM&CR, compliance failures are personal. Each check is tied to a defined rule, not an analyst's judgement call. If something goes wrong, you can show exactly what happened and why.
Regulation 40 Compliance
Reg 40 requires adequate audit trails for compliance decisions. Structured, timestamped records are generated automatically as part of each check, not bolted on after the fact.
How the Audit Trail Works
LangGraph FSM · Pydantic Schemas · Bayesian Trace
Each check runs through a finite state machine with typed data contracts validated by Pydantic. Name matching uses a Bayesian frequency engine calibrated against census data. You get full trace output: prior probabilities, likelihood ratios, posterior scores.
Why We Built This
85% of KYC/AML reviews at UK fintechs are still done by hand. Analysts copy-paste names into screening tools, eyeball the results, and make a call. There's no record of why. When the FCA asks, the answer is "an analyst reviewed it." That's not good enough.
We built Veritas to fix that. Each check produces a record of what data went in, what rules ran, and why the decision came out the way it did. Not a confidence score. An audit trail you can actually reconstruct.
We're not asking you to trust us. We're asking you to test us. The Shadow Pilot runs alongside your existing process, no integration required, and shows you exactly where the two approaches disagree.
The Shadow Pilot
4 weeks, no integration. We run alongside your existing compliance process and you get a report showing where the two disagree.
NDA & Setup
Sign the NDA and send us your applicant data. We configure the system to match your compliance rules and data format.
Backtesting
We process your historical applicants and compare our decisions against your existing outcomes.
Live Shadowing
The system runs in parallel on live applicants, making real-time decisions alongside your human reviewers.
Discrepancy Report
You get a detailed report comparing automated decisions against human outcomes, with a full audit trail for each check.
Observe only. No blocking or approval signals touch your production systems during the pilot.
Pricing starts with the 4-week pilot. Get in touch and we'll walk you through it.
Technical Architecture
Not a black box. Same input, same output, every time. You can inspect and reconstruct any decision.
Deterministic LangGraph FSM
Each applicant goes through a finite state machine with defined states and transitions. Same input always gives the same output. No randomness, no model drift. All state transitions are logged.
Pydantic Compliance Schemas
All data contracts are typed and validated at runtime using Pydantic. Malformed or incomplete data gets rejected before it enters the decision pipeline, not after.
Bayesian Frequency Engine
Name matching uses Bayesian probabilistic scoring calibrated against census frequency data. Each match comes with full trace output: prior probabilities, likelihood ratios, and posterior scores. You can explain exactly why a name matched or didn't.
Dual Signal Comparison
The account holder name and the document name are compared as two independent signals. Single-signal providers miss discrepancies that this dual comparison catches.
Veritas vs. Black-Box Providers
Veritas
- Deterministic: same input, same output
- Full audit trail for each decision
- Documented rule engine with typed schemas
- Bayesian trace output you can show to the FCA
- Dual-signal name comparison
Black-Box Providers
- Probabilistic: outputs change with model updates
- No reconstructible decision trail
- Opaque model weights, unexplainable scores
- Can't explain why a decision was made
- Single-signal screening only
Book Your Shadow Pilot
Tell us about yourself and we'll set up your 4-week Shadow Pilot. No disruption to your production systems.
We Tested It. Here's What Happened.
We ran 207 real-world civilian profiles and a full set of adversarial identity mutations through the system. These are the results.
207/207
Legitimate Applicants Approved
Zero false positives
All 207 legitimate applicants were approved without manual intervention. No real customers blocked, no onboarding friction, no lost revenue from wrongful rejections.
100%
Adversarial Mutations Caught
Every attack category detected
Synthetic names, transposed characters, phonetic substitutions. All manipulated identity data was caught across every adversarial category tested.
Full
Audit Trail for Every Check
Reg 40 compliant records
Each check automatically produced a structured, timestamped audit record that meets FCA Regulation 40 requirements.
Why This Matters: 2025 FCA Enforcement Actions
UK regulators aren't slowing down. In 2025 alone, the FCA handed out over £104M in fines for compliance failures:
£21.1M
Monzo
Compliance scaling failure
£44.1M
Nationwide
Transaction monitoring failure
£39.3M
Barclays
AML deficiencies
Total: £104M+ in fines
All results from empirical testing against real-world datasets. Run a Shadow Pilot to verify them against your own data.