The problem
Solicitors handling affordability and irresponsible lending claims were reviewing credit reports manually. Each file required a fee earner to read the report, assess the case strength, and write a Letter of Claim from scratch. No consistency in output. No speed. No scale.
At volume — processing large batches of CSV files — this approach was breaking down. The time per file was significant. The consistency of output varied by fee earner. There was no way to track which cases had been assessed, which were borderline, and which had Letters of Claim ready.
The work was fundamentally pattern-matching: read a credit report, identify risk indicators, assess whether the case meets the threshold, output a letter. The right system could do this at volume without fee earner time per file.
What we built
An automated affordability assessment platform built around the firm’s specific assessment criteria — configurable scoring logic, not a one-size-fits-all model.
Credit report upload triggering automated parsing and structured data extraction
Configurable affordability scoring engine rating each case and outputting a case strength rating
Automated Letter of Claim generation from scored output — structured, consistent, compliant
CSV batch upload for volume processing
Phase 2: open banking integration via Splink — post-pull transaction categorisation using MCC codes
Layered fallback logic to flag risk transactions where MCC categorisation is ambiguous
Exception handling workflow for cases requiring fee earner review
On the scoring engine
The affordability scoring engine is configurable per client. Thresholds, weighting, and flag criteria are set by the firm and can be adjusted without a rebuild. This is not AI making legal decisions — it is automated pattern recognition against rules the firm has defined. Fee earners review exceptions and edge cases. Straightforward cases go straight to Letter of Claim output.
Phase 2: open banking integration
The next phase adds Splink open banking integration. Post-pull transaction data is categorised using MCC codes with layered fallback logic for ambiguous transactions. This adds transaction-level evidence to the affordability assessment — not just credit report signals, but spending pattern analysis that strengthens the case file.
What changed
Phase 1 is live. Credit reports are uploaded, parsed and scored automatically. Letters of Claim are generated per case from the scored output. Batch CSV processing handles volume without linear fee earner time.
Fee earner involvement has moved from per-file manual review to exception handling only. Straightforward cases — the majority of the volume — flow through to Letter of Claim output without human intervention. Borderline cases are flagged for review.
The consistency problem is solved. Every Letter of Claim is generated from the same template with the same structure. The firm’s compliance exposure from inconsistent output is removed.