Kiya.ai

Renamed around artificial intelligence, the estate would still work without any model: core banking, digital channels and a regulatory reporting bench. What the models do is real but narrow: an analytics offering, a conversational interface and machine learning inside anti money laundering and fraud surveillance, with agentic AI now named among the product lines.

Origination is sold as fully automated, from campaign management and data capture through credit scoring and fraud detection, with robotic process automation and bots beside it. No approval step, review requirement or exception route is described, and nothing says what a person must sign off.

Surveillance, fraud detection and scoring models come with no published accuracy, error rate or validation method. How any of them reaches a decision is not described either.

Jordan Ahli Bank is named against a 60 percent cut in process turnaround on digital lending and a 30 percent rise in lead conversion on the relationship product. Both are workflow results for the platform, so they show the software works without isolating what the models add.

One vendor runs hosted core banking, money laundering surveillance and a customer facing chat channel for the same institutions, which makes the boundary between those surfaces a live matter. Pooling across institutions, model training on customer data and what the chat interface keeps from a conversation are not stated.

A website privacy statement and a modern slavery policy are all that is published, with no retention schedule, deletion terms or tenant separation statement for a core platform serving many institutions. That same firm sells data protection and compliance software to the institutions whose deposit, loan and transaction records it holds.

ISO 27001:2022 sits on a quality certifications page beside ISO 9001:2015 and a CMMI Level 5 appraisal, editions given. Only the first concerns security, and its certificate number, certifying body and scope are not published.

Regulatory returns and third party provider onboarding run through the software, but those are duties customers discharge, not a position the firm holds. As a software and services vendor it states no license, registration or supervised standing of its own, the usual position for a technology supplier.

Borrowers at rural banks, cooperatives and microfinance lenders are exposed to proxy discrimination and poorly placed to contest a decline, and they are who the scoring tools serve. Fairness testing, disparate impact analysis and a governance framework for the firm's own models are all unpublished, as is any position on AI regulation.

Customers meet automated outcomes through a mobile app, a chat interface or a village kiosk: a frozen payment, a blocked onboarding or a declined loan. The firm's recourse position is unpublished, as are whether a customer learns a model was involved, how an outcome is challenged and how responsibility divides with the institution.

Conversational, analytics and agentic AI features run on models with no provider, family or version given. Named technology dependencies all sit on the integration side, none in the model layer.

The firm supplies the system of record itself, so lending, microfinance and channels attach to its own core. A switch for India's unified payments interface and an open finance gateway, with its authorization protocols named and provider onboarding built in, round out the integrations.

Cloud native and hosted core banking are sold as separate offerings, yet no cloud provider, region or data location is stated. Many customers sit under data localization rules, so the missing residency terms bear directly on them.

Every route ends at a demo request, with no tiers, module structure or licensing model published for software sold as cloud, hosted and on premises. Whether AI features carry a charge separate from the core and regulatory modules is not said.

More than 450 financial institutions in more than 55 countries are claimed, from commercial and cooperative banks to microfinance institutions and governments, and microfinance and Islamic banking get their own product modules. Client counts differ between the firm's pages and its press boilerplate, and cooperative, rural and commercial banks share one set of general material.