Finvero

The removal test leaves a directory of lenders and manual underwriting. Alternative credit scoring built on generative techniques and alternative data produces decisions in under five minutes for applicants without conventional files, a fraud engine runs alongside it, and a separate predictive model ranks delinquent customers by likelihood of repayment. The company describes online models that improve continuously from origination data and from analysis of loans that went unpaid, so the scoring layer is the product rather than a feature of it.

Policy stays with the lender by design rather than by assurance. Institutions build their own scoring models on the platform, choosing between traditional, machine learning and predictive approaches, configure their own evaluation criteria and dynamic application flows, and select their own fraud rules, so the vendor supplies the engine while the institution owns the decision rules. Held at B because no threshold, escalation or human review requirement is described, and a five minute decision implies most applications complete without anyone looking at them.

One disclosure is genuinely rare and lifts this above its peers: the collections model's full feature set is published, naming payment history, frequency of arrears, days in delinquency, partial payments, indebtedness level, income to active debt ratio, credit age, time since loan opening and payment channel. Publishing the inputs of a scoring model makes the method inspectable rather than asking for trust, which is what supervisory guidance asks of banks buying third party models. A bounded effectiveness figure accompanies it at 10 to 15 percent improved decision accuracy. Held at B because no validation methodology, baseline or error rate is given for that figure.

One named partnership carries most of the weight: the company is described by a card network's inclusive growth programme as a strategic ally supporting micro-entrepreneurs, with that partner stating the platform gives its participants access to better credit terms. A published figure claims lenders improve decision accuracy by 10 to 15 percent. Beyond that, no lender, merchant or institution is named, no origination volume or customer count appears, and the merchant and lender network is described as national without any figure attached.

The company states the mechanism itself, which makes the absence of a boundary more pointed rather than less. Models are described as improving continuously by learning from data generated during origination and from loans already granted, and specifically as extracting value from user information by analysing unpaid credits. On a multi-lender marketplace that means defaults observed at one lender improve the models scoring applicants for its competitors. Nothing states whether that pooling is disclosed to participating lenders, whether it can be declined, or what a lender contributes by joining.

No data protection agreement, retention schedule, subprocessor list or consent framework was located. The platform holds identity, income, debt and repayment data on consumers and small businesses across two countries, both of which have their own personal data protection statutes, and it draws on alternative data whose collection basis is not described.

No attestation, certification, trust centre or enumerated framework was located. Regulated finance companies connecting origination, servicing and collections through the platform would require security assurance before doing so, and nothing is published for a prospective lender to assess before engaging.

Named regulated entity types anchor this, with the platform explicitly serving popular finance companies and multiple purpose finance companies, which are the licensed non-bank lender categories in its home market and carry distinct supervisory obligations. Regulatory compliance is named as a design requirement of the collections module specifically, which is the correct emphasis given debt collection conduct rules. Identity validation covers both individuals and companies. Held at B because no regulator or statute is named for either country of operation.

The inclusion evidence is external rather than self declared, which is what earns the grade: a card network's inclusive growth programme names the company a strategic ally for micro-entrepreneurs and states that its participants access better credit terms as a result, and alternative scoring extends assessment to applicants with no conventional file. The counterweight is specific and visible because the company publishes its features.

Its collections model uses payment channel as an input, distinguishing whether someone pays at a branch, online or in cash, and paying in cash at a branch is a proxy for being unbanked or lower income. That signal then feeds a model determining how hard a delinquent customer is pursued. No fairness testing appears.

No guarantee, indemnity or correction process was located. The borrower is affected at both ends and addressed at neither: declined in under five minutes on an alternative score they cannot see, or segmented by a collections model into a recovery priority based partly on how they choose to pay. Nothing describes whether reasons are given, how incorrect alternative data is corrected, or how someone contests a delinquency classification.

Inputs are described only as alternative data and the volumes analysed, with no bureau, telecommunications provider, open banking aggregator or data vendor named, and no model provider identified behind the generative capability. For a scoring platform whose distinguishing claim is alternative data, the provenance and permissions of that data are the disclosure a buyer and a regulator would both examine.

The platform covers the operational spine a lender needs rather than one stage, including loan servicing, collections, reconciliations and returns, identity validation for individuals and companies, and access to an established merchant and lender network across the country. Interfaces and open banking connectivity are referenced. No named core banking, servicing or bureau system appears, and no developer documentation was located.

No hosting provider, region selection, residency commitment or private deployment option was located. Operating across two Latin American markets whose data protection regimes differ, while holding consumer credit and identity records, makes processing location a question a regulated non-bank lender would raise during procurement.

No pricing, packaging or basis of charge was located. Trial framing indicates no installation or card required to begin, which describes onboarding rather than cost, and for a marketplace the material question is whether the platform takes a fee per origination, a share of interest, or a licence, since that determines its incentives on approval volume.

Buyer coverage names the specific regulated entity types that matter in this market, spanning banks and financial institutions, popular finance companies, multiple purpose finance companies, fintechs, merchants and businesses of all sizes, which is more precise than a generic list.

Both consumer and business lending are supported, across instalment, revolving and deferred payment products including in-store origination, and the platform covers origination, risk, collections and portfolio management. Operations span two countries with stated regional expansion.