AIZEN Global

Both businesses are models. The automated machine learning platform exists to let institutions build and operate predictive models across underwriting, fraud detection, anti money laundering, auditing, claims and marketing, and the lending platform converts non financial platform data into credit decisions that could not otherwise be made, since the merchants and drivers being assessed have no conventional credit basis. Explainable techniques are applied specifically to small business creditworthiness, and a deep learning fraud system processes over 15,000 transactions per minute with near real time retraining.

The stated ambition is the most far reaching automation claim in this index and no counterweight accompanies it. The lending platform is described as being designed and built as an autonomous banking operating system that fully automates core credit decisions across customer acquisition, product development, risk management and collection, which is the entire credit lifecycle with no human checkpoint named at any stage.

Collections is the most consequential inclusion, since automated pursuit of borrowers in difficulty is where oversight matters most. No escalation route, threshold or review requirement is described, and the modelling platform's own premise that users can build predictive models without expertise in artificial intelligence compounds the question of who is supervising what.

The model management tooling is the most substantial described by any vendor here: a systematic framework to build, monitor and update thousands of predictive models in parallel within a single view, with visualisation across data, features, model performance and prediction results. That is model risk infrastructure rather than a claim about accuracy, and it addresses the real problem for an institution running many models, which is knowing which are drifting.

Explainable output supports credit decisions specifically, one published figure gives over 95 percent accuracy on product recommendation, and the fraud system retrains almost in real time. Held at B because no accuracy or validation figure is published for the credit models themselves, and because the platform's premise that users without artificial intelligence expertise can build and deploy predictive models is a governance risk the company presents as a feature.

The platform is described as used by some of the largest banks, credit card and insurance companies, with major card issuers specifically adopting the fraud detection system. Market entry evidence is unusually concrete: 117 affiliated companies signed within eight months of launching in Vietnam, partnerships with six local financial authorities, and a named mobility partnership with a major regional platform.

The award record is deep and independent, including second globally in a central bank sponsored fintech award alongside a global reinsurer and a global custodian, first place at a major Asian fintech week, a top ten listing by an international bank, and an analyst firm's emerging vendor recognition. Series B funding came from a state led innovation fund backed by major domestic banks.

No boundary statement was located. The stated strategy is vertical data economy integration, meaning platform data flows into models that banks then rely on, and the company acquired a data aggregation business to reinforce that flow. Nothing states whether a merchant's sales and returns data informs models scoring competitors on the same marketplace, what a platform partner surrenders by integrating, or how long behavioural data persists after a loan closes.

No data protection agreement, retention schedule, subprocessor list or consent framework was located. The model depends on moving non financial data into credit decisions, drawing on commerce, mobility, education and healthcare platforms, and one described product derives a financial health score from digital communication data. Health platform data used for lending decisions raises questions in every jurisdiction served, and Korea's own data protection regime is among the stricter ones, none of which is addressed publicly.

No attestation, certification, trust centre or enumerated framework was located. Large banks, card issuers and insurers run the modelling platform and a national regulator has designated the company for a regulated function, so security assessment has been conducted repeatedly at a demanding level, and none of the resulting control documentation is published for institutions in newer markets to rely on.

This is the strongest regulatory position in the index. The company states it was the first designated by Korea's Financial Services Commission to underwrite loans on behalf of banks using AI-driven credit decisions, which is not a sandbox admission or a compliance claim but a national regulator formally authorising a specific company to perform a regulated function by machine, and being first means the designation was created around this approach.

Market entry abroad was conducted with regulators rather than around them, with six local financial authorities partnered in one country. The company also appears on an independent listing of firms recognised for responsible artificial intelligence alongside major payment, analytics, bureau and audit firms.

Explainability is applied where it matters most, with explainable techniques named specifically for assessing creditworthiness of micro and small business owners from a multi dimensional view, and the stated outcome is two sided, with banking partners acquiring more clients while offering better rates and higher limits rather than simply approving more. Independent recognition for responsible artificial intelligence alongside established payment and bureau firms adds outside assessment.

Held at B because one input set carries the proxy risk recorded elsewhere in this index, namely deriving a financial health score from digital communication data, which correlates with wealth and social position, and because no fairness testing, approval rate analysis or subgroup outcome data is published.

No guarantee, indemnity or correction process was located, and the automation ambition makes the absence more consequential. A merchant or driver assessed on sales data, return rates and platform behaviour receives a credit limit and interest rate set by model, and if the system operates as intended across acquisition through collection, no human is stated to be involved at any point they could appeal to. Explainable output serves the bank rather than the borrower, and nothing describes whether reasons are given, how a wrong platform data point is corrected, or how a decision is contested.

The core dependency is internal and named, since the modelling platform is proprietary and purpose built for financial services, which places the company among the minority here that own their modelling stack rather than wrapping a third party interface, and it acquired its data aggregation capability rather than licensing it.

What is not disclosed is the platform data side: the commerce, mobility, e-wallet, education and healthcare sources feeding credit decisions are described by sector only, with no partner, licensing arrangement or subprocessor list identified.

Integration is designed to be non disruptive and the company says so precisely, exposing platform functions through interfaces that allow the existing system to remain unchanged, which is the condition under which a bank will adopt anything touching credit.

The harder integration achievement is on the other side, connecting bank lending infrastructure to commerce, mobility, e-wallet, education and healthcare platforms whose data was never structured for credit use, supported by acquiring a data aggregation business outright. No named core banking or origination system appears.

No hosting provider, region selection, residency commitment or private deployment option was located. Operations span at least five jurisdictions including markets with financial data localisation requirements, and the platform holds behavioural and transaction data from consumer platforms alongside bank credit information, which makes where processing occurs a question a domestic regulator would ask before permitting the arrangement.

No pricing, packaging or basis of charge was located across two distinct products with very different economics, a modelling platform licensed to institutions and a lending platform embedded into consumer journeys. Nothing indicates whether the lending side is compensated per origination, by revenue share with the financing partner, or by platform fee.

Institutional coverage spans banks, credit card companies and insurers, and the platform side reaches non financial sectors that rarely appear together, including e-commerce, mobility and electric vehicles, e-wallets, education, healthcare, retail and the gig economy. Functional breadth is equally wide, covering underwriting, fraud detection, anti money laundering, auditing, claims analysis, product development, marketing response and collections. Geographically it operates across Korea, Vietnam, Indonesia, Singapore and Hong Kong, with market entry evidenced rather than asserted in at least two of those.