CareEdge Analytics

The Clearwater precedent, and the fifth consecutive centrality C off this analyst roster. The suite is regulatory calculation and workflow software: Basel II and III compliance, asset liability management, fund transfer pricing, IFRS 9 expected credit loss provisioning, loan origination, market and operational risk. Strip every learned component and a complete working risk and regulatory reporting platform remains, because that is what banks have run on it since 2006.

The vendor settles the question against itself in the same way Loxon did: the credit risk assessment page states that lenders can choose to use their own proprietary models or leverage the vendor's credit scoring models. A vendor that tells the buyer the model is optional is telling the buyer the model is not the product. The generative platform branded EdgeAvira.ai and the graph based analytics work are real but sit as lines beside a large conventional suite.

Oversight is asserted and partially described but never placed in the execution path. The credit risk assessment product is documented as blending objective and subjective methodologies, which puts a credit officer's judgement alongside the model output, and the impairment workflow is described as repeatable, auditable and consistent with visual workflow management.

The buyer also retains model control, since lenders may run their own proprietary models instead of the vendor's scoring models. What is absent is the enforcement layer: no confidence threshold, no escalation route, no stated review band, and nothing describing what happens at an automated decline in the origination flow, which is the point in a lending workflow where oversight matters most.

Methodology is described in places but never validated. The expected credit loss material sets out standard and custom macroeconomic scenario generation for the forward looking and probability weighted requirements of IFRS 9, and describes an auditable repeatable process, but that restates the accounting standard rather than evidencing model quality.

No backtesting, no discriminatory power measurement, no calibration or drift monitoring, no versioning policy, no independent validation, and no accuracy figure for the generative risk intelligence platform. This falls short of the Loxon bar, where an early warning product page carried backtesting, signal significance testing, reject inference, plausibility checking and rating migration matrices in the supervisor's own vocabulary.

Queued check, and it is the same one banked against CRIF: a registered credit rating agency is obliged by that registration to publish its rating methodologies and disclose material changes to them, so externally compelled model documentation probably exists on the ratings side and should be examined.

Two named bank customers give testimonials on the product site, and both name the specific modules deployed rather than offering generic praise. Mashreq Bank of the United Arab Emirates records the completion of an operational implementation, and Union Bank of Colombo in Sri Lanka records the rollout of the credit, operations and market risk modules and credits the vendor's consultants with delivery and staff training.

A database vendor separately published a case study describing the fund transfer pricing application in production at one of the software arm's largest customers, a large public sector bank producing regulatory profitability reports, though that bank is not named. The parent group is placed in an independent risk technology ranking.

Held at B because no quantified outcome is attached to any named institution: the two named customers give qualitative testimonials with no figures, and the one story carrying operational detail leaves the bank unnamed.

Nothing states whether customer data trains the vendor's models. This matters more here than for most suppliers because of the group structure: the same group operates a credit rating agency, a business analytics arm and a bank software arm, so a bank buyer has a direct interest in whether its borrower data, exposure data and portfolio performance can inform anything on the ratings or research side.

No separation commitment, no exclusion of client data from any training corpus, and no statement of controls around the generative platform are published. Against the reference set of Mortgage Capital Trading, Needl and AlphaSense, all of which answer this question plainly, silence is a choice.

No privacy posture is published. Nothing addresses retention periods for borrower financial data, purge on contract termination, data subject rights under India's Digital Personal Data Protection Act, or the handling of the borrower financial statements and transaction records that the spreading and fund transfer pricing applications necessarily ingest.

No security certification is claimed in any material reviewed. No ISO 27001, no SOC report of either type, no penetration testing statement, no trust portal and no security page. This is a supplier handling loan level and transaction level data for banks in several jurisdictions, so the absence is a disclosure choice rather than a category constraint. Queued check: the vendor site disallows automated access, so a manual pass over its footer and policy pages could move this grade.

Two distinct forms of standing, in four jurisdictions, and the second is the one that carries weight beyond a compliance claim. The parent is a credit rating agency registered with the Securities and Exchange Board of India and recognised by the Reserve Bank of India, meaning it is directly supervised, investigable, sanctionable and capable of being deregistered, which is a position no ordinary software supplier in this index occupies.

Separately, CARE Ratings Africa is licensed by the Financial Services Commission of Mauritius and recognised as an External Credit Assessment Institution by the Bank of Mauritius, which makes its assessments usable by supervised banks for regulatory capital purposes. Rating entities in Nepal and South Africa and a company in the GIFT City international financial services centre sit alongside.

Recorded limitation: the authorisations attach to the ratings parent and its rating subsidiaries, not to the software entity that sells the Kalypto products, so the standing is group level rather than product level.

No fairness position of any kind. No disparate impact testing, no protected characteristic handling, no model inventory, no named governance framework, and no reference to any AI governance standard. The specific exposure is concrete rather than theoretical: the products generate credit scores and risk grades that drive lending decisions for banks and non banking financial companies across India and South Asia, including retail and small business borrowers, and the group's own credit scoring models are offered as an alternative to the lender's. A scoring model supplied by the vendor and applied across many lenders concentrates any systematic bias it carries.

No recourse position is published. Nothing states what happens when a risk grade, an impairment calculation or an origination decision is wrong, whether the lender or the vendor bears the consequence, or whether a borrower is told that a model contributed to a decline.

The lending exposure follows the standing pattern for software suppliers in this index: the bank is the regulated party answerable to its supervisor and its customer, while the supplier that built the scoring model sits outside the conduct perimeter, even though the parent group is itself supervised in its rating capacity.

Three technology dependencies are nameable and none is named by the vendor in its own product material. A graph database provider announced in its own press release that it supplies the graph analytics behind the vendor's customer and fraud work, including anti money laundering toolkits. A database platform vendor published a customer story describing the engine underneath the fund transfer pricing application.

The parent's own corporate disclosure records a memorandum of understanding with an analytics firm to launch predictive intelligence products in India. Most materially, no model, family, version or provider is named anywhere for the generative risk intelligence platform, which is the one product where the dependency question actually bites.

Graded C on the Provenir precedent: where the disclosure lives in supplier announcements and trade coverage rather than in vendor material, the vendor has chosen generic language over a nameable dependency. This is also a third confirmation of the standing sourcing tell that the supplier discloses the vendor.

A wide functional footprint across the bank: enterprise risk management, asset liability management, fund transfer pricing, IFRS and financial reporting, lending origination, early warning signals, collections and recovery, and credit ratings and scoring.

The loan origination system is documented as omnichannel across online, mobile, social and branch touchpoints with straight through transaction processing, and the fund transfer pricing application is documented as evaluating every individual customer transaction at scale.

Held at B rather than A because no core banking platform is named as a certified integration, no connector catalogue is published, and integration depth has to be inferred from the functional breadth rather than read off a stated list.

No deployment or residency position is published. Nothing states whether the Kalypto products are delivered on premise, in a managed cloud or as software as a service, which cloud regions are available, where customer data is processed and stored, or how the arrangement satisfies Reserve Bank of India data localisation requirements. That last omission is notable given the Indian bank buyer base, since localisation is a live procurement question for exactly those institutions.

No pricing information of any kind was found. No published price, no tier structure, no unit or per seat basis, no indicative implementation cost, no minimum volume, and no statement of the commercial model for either the licensed software or the advisory services sold alongside it. Every route into the product is a contact form. The product site also blocks automated access, so the assessment rests on indexed pages and third party sources.

Broad across institution type and geography. The software arm is described in a partner announcement as serving four thousand customers worldwide with risk management, finance automation and regulatory and financial compliance software, and it addresses the full banking, financial services and insurance segment rather than one product line.

Named bank users span three markets: a large Indian public sector bank running the fund transfer pricing application for government profitability reporting, Mashreq Bank in the United Arab Emirates, and Union Bank of Colombo in Sri Lanka. The parent group operates rating businesses in India, Nepal, Mauritius and South Africa plus a GIFT City entity, and has completed more than ninety one thousand rating assignments covering banks and financial services institutions. The four thousand figure originates with a technology partner rather than the vendor and is recorded as such.