JurisTech

Models do identifiable work, predicting which delinquent accounts will self cure and which are heading to non performing status through behavioural scoring and predictive and prescriptive analytics, with three named intelligence products layered over the platform and generative applications described across risk assessment, loan structuring and recovery.

Held at B because the underlying business is enterprise lending software, and the company says so itself with unusual candour, observing that in fast growing credit markets the priority is often digitalisation before AI transformation. The named bank deployment describes configurable rule based scoring alongside the AI framework, and stripping the models leaves an origination and collections platform half a country's banks already run.

Institutions retain control of strategy rather than receiving decisions, configuring collection approaches themselves and testing them through champion challenger comparison with feedback, so the platform executes a policy the bank designed and measured rather than one the vendor set. Treatment tracks are selected by the institution against predicted delinquency behaviour.

The multi party design keeps humans central by connecting lenders, agencies and solicitors on a shared platform rather than automating them away. What is absent is any description of thresholds, of what proceeds without review, or of oversight over the prediction that assigns a borrower to a track.

Champion challenger testing is built into the collections product rather than offered as a service, letting an institution run competing strategies against each other and determine empirically which performs best for which customer, with feedback closing the loop. That is the same experimental design that produced the strongest validation evidence elsewhere in this index, here shipped as a feature the customer operates. Auditability is named as a property that must be embedded. What is missing is measurement of the predictions themselves: no accuracy for self cure forecasting, no early warning precision and no validation result is published.

The penetration claim is in the same class as the strongest in this index: more than half the banks operating in Malaysia use the platform. One deployment is named and quantified four ways, with a bank reporting 30 percent faster application turnaround, 30 percent higher staff productivity, personal loan disbursement cut from three days to one, mortgage processing from seven days to three, and same day credit card approval enabled.

A major financial publication named it Asia's best lending solution for 2026. The country's leading credit bureau is a corporate investor, and the company acts as channel partner on a named consumer finance deployment. Expansion into two further Southeast Asian markets is under way.

No boundary statement was located and the concentration makes the question unusually pointed. Serving more than half the banks in a single market means behavioural scoring, self cure prediction and collections strategy models are trained on and applied across institutions competing directly for the same borrowers.

Nothing states whether learning is tenant isolated, whether one bank's repayment outcomes improve models scoring another bank's customers, or what a lender can decline to contribute. The corporate investor being the national credit bureau adds a further unaddressed relationship.

No data protection agreement, retention schedule, subprocessor list or deletion commitment was located. The platform holds origination files, credit decisions, bureau records drawn from both the central registry and the private bureau, and collections histories for borrowers across most of one country's banking system, which is a concentration of personal financial data few vendors anywhere match. Malaysia has its own data protection regime and none of it is addressed publicly.

No attestation, certification, trust centre or enumerated framework was located, with security named as a principle that must be embedded alongside policies, transparency and auditability. More than half a country's banks have completed supplier assessment on this vendor, so the underlying controls have been examined repeatedly, and none of that assurance is published for institutions in the newer markets to rely on.

Two named external systems anchor this, integration with the central bank's own credit reference registry and with the national private bureau, which means the platform operates inside the country's formal credit information infrastructure rather than alongside it. The company also states the principle directly, that policies, transparency, auditability and security have to be embedded rather than added, which is the correct framing for software half a banking system depends on. Held at B because no regulator, statute or specific supervisory requirement is named, and expansion into two further jurisdictions brings regimes that go unmentioned.

One capability distinguishes this from every other collections vendor in the index and deserves the grade on its own. The platform predicts self curing accounts, meaning borrowers who will resolve their arrears without any intervention, which allows an institution to leave them alone.

That is a direct reduction in unnecessary pressure on people in temporary difficulty, and it inverts the usual logic of the category, where the models exist to decide how hard to push rather than whom not to push at all. The company also describes empathetic recovery strategies and better customer service as stated objectives. Held at B because behavioural scoring still segments people into treatment tracks, and no fairness testing, outcome analysis or track assignment audit is published.

No guarantee, indemnity or correction process was located. The institution is well served by champion challenger measurement and its own control over strategy. The borrower has nothing described and is affected at several points: a self cure prediction that proves wrong means unnecessary contact, an early warning flag can tighten terms before any default occurs, and a delinquency track assignment determines how they are treated. Nothing states whether reasons are given, how bureau data errors are corrected, or how a borrower contests the classification driving their treatment.

The two external credit data dependencies are named individually, the central bank registry and the national bureau, which is the disclosure that matters most for a lending platform because those sources determine what any decision can see and the second of them is also a corporate investor in the company. Three internal intelligence products are named as the analytical layer. What is not disclosed is any model provider behind the generative capabilities described, any hosting arrangement, or a subprocessor list.

Two external systems are named specifically and they are the ones that matter in this market, the central bank's credit reference registry and the national private bureau, without which no lending decision in the country is complete. The platform also acts as connective tissue between institutions, collection agencies and solicitors in real time, which is integration across organisations rather than only across software. Architecture was modernised in 2025 with the platform moved from virtual machine environments to containerised cloud native deployment. No core banking system is named.

No hosting provider, region selection or residency commitment was located. The architectural disclosure is more specific than most, with containerised cloud native deployment completed across the platform in 2025 replacing virtual machine environments, which indicates deployment flexibility including potentially within a bank's own environment, and the company does not say where customer data actually sits or whether on premise remains available.

No pricing, packaging or basis of charge was located across a suite spanning origination, decisioning, collections and analytics products. The value case is made through operational outcomes at a named bank rather than through cost, and nothing indicates whether licensing is per module, per institution, per loan or by volume, which for an enterprise platform sold to banks of very different sizes is the material question.

Buyer coverage is genuinely wide and includes types most vendors here do not reach, spanning banks and digital banks, multi finance companies, leasing firms, government backed lenders, non bank financial institutions, collection agencies and solicitors, plus telecommunications companies.

Functional coverage runs the entire credit lifecycle from onboarding and origination through decisioning and administration to early warning and recovery, which is why a single vendor can serve half a banking system. Geography is the constraint: this is Malaysia with expansion into Indonesia and the Philippines, so regional rather than global.