Live demo
FinLend
- Load one of the synthetic borrower packs, or drop your own files.
- Read the risk score and what each factor did to it.
- Ask the on-prem assistant what would change the decision.
- Email-code sign-in
- Explainable risk · SHAP
- On-prem AI assistant
- OFAC screening · on-prem
How the demo works
- 1
Load a document pack and run the mapping.
- 2
Check the parse outcomes, the coverage and the score.
- 3
Ask the assistant about the file.
What you are looking at
Three things built on top of the FinLend domain data model — all running on your own infrastructure.
Email-code sign-in & user management
Passwordless one-time-code sign-in. Admins invite teammates by email and control who sees what, so a whole underwriting team works in one place.
Explainable gradient-boosting risk score
A gradient-boosting model scores each borrower and SHAP surfaces the factors that pushed the score up or down — no black box. Trained on synthetic (but realistic) data for the demo; a ~30-minute retrain swaps in your real book for production.
On-prem AI assistant over the borrower's docs
A fully on-premise assistant answers from the borrower's own documents and the risk result — including counterfactuals like "what would need to change for a positive decision?" Nothing ever leaves your infrastructure.
Compliance & data checks
Screening runs on your own infrastructure today; every other check plugs into the same borrower file through your existing provider — examples below.
Sanctions & watchlist screening
Borrower, guarantor and entity names are matched on-box — no data leaves your network. EU, UN and UK/HMT lists drop in the same way.
PEP & adverse-media screening
Flag politically-exposed persons and negative news against the borrower and its principals.
Credit & liabilities
Pull bureau scores, tradelines and outstanding liabilities to validate the borrower's real debt load.
Business & identity verification
Confirm the entity, officers and good standing (KYB) and verify the people behind it (KYC).
Bankruptcy & litigation
Surface filings, judgments and liens tied to the borrower or its guarantors.
Provenance, audit trail & fraud signals
Every extracted figure is traced to its source document, with balance-sheet consistency, cross-document agreement and PDF-metadata / Benford anomaly checks — all computed on your box.
Tell us where
it breaks
Bring the part of your fintech stack that is hard right now. We start with a working consultation and scope to outcomes.