Helping SMEs understand and improve their fundability

A FinTech business working with Cloud Combinator on AWS. The client is anonymised at their request.

The work centred on three focus areas that together turn a raw score into something a SME can use.

On the diagnostics side, Cloud Combinator built and compared several models, including Random Forest and Gradient Boosting classifiers, with a Causal Forest explored to estimate how specific improvements would change a SME's position. Model selection is evidence-led: the best performer is used, or predictions are combined where that helps, or the more interpretable option is chosen where performance is comparable. Amazon SageMaker underpins training, monitoring and inference, and SageMaker Clarify with SHAP provides feature-level explanations and bias detection throughout.

The engagement delivered an explainable, bias-aware diagnostics model and a LLM-powered remedy layer against the success criteria in the Statement of Work, with acceptance measured by expert panel review of plausibility and usefulness versus the existing pilot. The figures below describe the intended operating scale and unit economics rather than an approved production benchmark.

• Richer fundability diagnostics — Forest-based models assess a SME's position on the fundability spectrum and the impact of changing key variables, going beyond the rule-based pilot while preserving the client's segmentation.
• Transparent and bias-aware — SHAP explanations quantify how each factor affects a score, and SageMaker Clarify monitors for bias across groups, with corrections applied or documented where material.
• Actionable, tailored remedies — The Bedrock layer turns technical outputs into clear, bank-safe steps a SME can take, with language and depth adapted to whether the business is a startup or an experienced trader.
• Compliance-conscious guidance — Remedy text is designed to pass internal compliance review and to assess eligibility and fundability without making explicit lending or borrowing-amount decisions.
• Handover — The client received the models, remedy layer and AWS architecture with documentation, runbooks and knowledge-transfer sessions for continued development in their own environment.

With diagnostics and remedies working together, the client can help a large annual population of SMEs not only see where they stand but steadily improve their fundability, deepening engagement well beyond a single loan quote.