A lease application from a business customer arrives with attachments: a balance sheet and profit and loss statement for two years, several months of bank statements, the seller's quote with the vehicle VIN or equipment serial number, sometimes the shareholder structure. The risk analyst opens every file, retypes the figures into the Soft4Leasing or Business Central application record or an Excel scoring sheet, separately requests an Experian report and a business register extract, and checks whether the income in the statements agrees with the financial report, whether the asset price in the quote matches the application, and whether the income on the social security certificate matches an individual's application. When a document is missing, the customer finds out days later, and the seller waits for an answer in the meantime.
The solution receives the application and attachments from e-mail, a partner portal or the self-service portal. Document recognition and a private model, running in the MDP CLOUD GPU cloud in Lithuania (EU) or on Amazon Bedrock in an EU region, as you choose, identify each document type, extract figures, amounts, dates and the VIN or serial number, and write them into the Soft4Leasing or Business Central application record as a draft marked "awaiting approval". It then pulls the Experian report and the business register extract through the access you already have, cross-checks amounts between documents, and flags mismatches, missing documents, outdated periods and debts reported by Experian. The analyst receives a prepared record with notes and source links and makes the credit decision personally: the model does not assess risk and does not make recommendations. We connect to Soft4Leasing and Business Central through an API, a file export or an e-mail flow, depending on the version. Data does not leave the EU, and every processing step is logged in line with GDPR and DORA.