C&R Software

The Clearwater precedent, and the vendor's own definition settles it against itself. Its buyer guide defines an AI native collections platform as one where machine learning, predictive analytics, automation and decisioning were designed into the core architecture from the beginning rather than layered onto legacy infrastructure as aftermarket additions, and it applies that label to a platform it elsewhere describes as having forty years of history under three owners.

A system of record that ran enterprise collections for decades cannot have had machine learning in its core architecture from the beginning. Strip the models and the delinquency workflow, account staging, agency placement, legal and bankruptcy handling all continue, because they did for most of the product's life.

Gates are asserted and partially located. The agentic framework is described as delivering governed and explainable AI, the platform is positioned for teams using AI to support human judgement rather than replace it, and the operator interface adjusts what it presents according to customer situation, account status and applicable regulatory requirements, which places a compliance constraint at the point of the collector's interaction.

Held at B rather than the Skit precedent A because no rule is cited: the interface is said to respond to regulatory requirements without naming a single one, where the reference vendor in this same pocket encodes calling windows, contact frequency caps and prohibited contact lists against a named regulation. Champion and challenger testing provides a described route for changing strategy under control.

A weak B resting on two real properties rather than on disclosure. FitLogic is a decision engine that lets the institution build, deploy and update its own machine learning models inside the collections environment, and champion and challenger testing is named as the mechanism for validating a strategy change against the incumbent before it is adopted, which is a genuine and supervisable control. Both help the buyer manage models it owns.

Nothing at all is published about the vendor's own layer: the agentic framework, the chatbot and the interface logic carry no documentation, no evaluation, no versioning and no drift statement, so this repeats the pattern where a platform selling model tooling discloses nothing about its own models.

The most striking gap on this vendor and worth stating plainly: the largest installed base in its category publishes no checkable customer evidence. Testimonials carry precise job titles attached to deliberately unnamed institutions, a senior manager for consumer financial wellbeing and a vice president for strategy, analytics and data governance, both at a top ten United Kingdom bank, and the scale claims follow the same pattern: five of the ten largest United Kingdom banks, more than twenty million accounts managed, over two hundred and eighty organisations, more than sixty countries.

Not one institution is named by the vendor and no outcome is attributed. A third party install base database lists several institutions as users, but that is a commercial dataset rather than a vendor claim and is not credited here. Under the standing bar this is no named customer and self reported figures only.

Unanswered. Nothing states whether account, payment behaviour or conversation data from one institution informs models available to another, which is the live question for a platform holding more than twenty million accounts across hundreds of organisations in one hosted environment.

Building the agentic framework on a managed foundation model service does carry an implication about where data goes, since that service does not by default use customer content to train the underlying models, but the vendor never makes that commitment in its own words and inference is not disclosure.

Little beyond what the security certifications imply. Partner collaboration is described as maintaining complete control over sensitive information, which is an access statement rather than a privacy position, and nothing published sets out retention periods, deletion rights, subprocessors or cross border transfer mechanisms.

For a platform holding delinquency, hardship and legal action records on consumers across more than sixty countries, and reaching those consumers directly through messaging and a chatbot, the absence of a published data handling position is material.

The most precisely enumerated credential set encountered this session. Three certifications are named and each carries the qualifying detail that usually goes missing: payment card industry data security standard at level one rather than an unspecified level, service organisation control at type two rather than an unspecified type, and information security management at the 2022 edition rather than an undated reference.

That precision is exactly what separated this from two weaker presentations graded the same week. Held at B rather than A because no certificate numbers, accrediting or attesting bodies, examination periods or trust portal are published, and the certifications appear on a product page rather than in a security centre.

A software vendor with no licence, registration or supervised standing. This is the structurally important half of the pocket's liability finding: unlike a licensed collection agency operating in the same market, this firm sits outside the perimeter entirely and cannot be sanctioned by a conduct regulator for how an account is worked, however much of the working its software determines.

No fairness position published. Explainable and governed are used as adjectives for the agentic framework without a method behind either, and nothing addresses differential treatment across customer groups in a product whose entire function is deciding which people to contact, how often, through which channel and on what settlement terms. Contact intensity and settlement offers are precisely the decisions where disparate outcomes arise. A hardship route to a directory of vetted support resources is a real consumer facing provision and is credited as such, but it is a service feature rather than a governance disclosure.

Confirms the pocket finding at the top of the market, which is why this vendor was worth building. Every software vendor in the collections pocket takes C here because the creditor absorbs the statutory consequence of a model it did not build and cannot fully validate, while a licensed agency is itself the regulated party and can be complained about and sanctioned directly.

This is the largest software vendor in that pocket by installed base and it follows the pattern exactly: no statement of who answers for a wrongly prioritised account, an inappropriate contact sequence or a settlement decision, and no route for the consumer to contest one. The hardship resource directory gives the consumer something, but it is assistance rather than recourse.

Rare in this index and rarer still in this pocket: the vendor names the layer its agentic framework is built on, a specific managed foundation model service from a named cloud provider, in its own material rather than leaving it to trade coverage. That tells a buyer the perimeter its prompts and account data operate within and which supplier relationship sits behind the capability, which is materially more than the generic large language model reference most of this cohort offers.

Held at B rather than A because no individual model, family or version is identified, no version policy is stated, and there is no per component breakdown of which capabilities use the service and which use the vendor's own predictive models.

A fuller integration story than most of this cohort. The product is positioned explicitly as a single system of record consolidating fragmented collections estates, with an open architecture, prebuilt connectors for major core banking platforms, a choice of real time interfaces or batch processes, and a configurable rather than customisable model intended to avoid bespoke code. An integration time reduction of up to seventy five percent is claimed through configuration. Held at B because the core banking connectors are asserted as a category and no platform is named, which is the same gap that held three other vendors this session below an A on this axis.

Above the cohort norm because the hosting arrangement is described rather than implied. A hosted service runs in the company's own cloud built on named public cloud infrastructure, with a multi region design, always on backups and disaster recovery capability stated, alongside a traditional on premises deployment retained as a separate option, and planned update cycles for features, security and compliance.

Held at B because no specific regions or countries are enumerated and no residency commitment is made, which matters for a platform operating in more than sixty jurisdictions where collections data is among the most locally regulated categories there is.

No pricing published: no rates, tiers, per account or per seat figures, and no indication of how the hosted service is charged against the on premises alternative. Every route ends in a selection conversation. The only commercial datum published anywhere is a statement that more than ten million dollars has been invested in the hosted platform, which describes the vendor's spending rather than the buyer's.

Among the broadest in the index. Stated presence at five of the ten largest United Kingdom banks, top tier banks in the United Kingdom and United States, operations in more than sixty countries, over two hundred and eighty organisations on the hosted service and support for more than six hundred and fifty distinct debt types.

Financial buyers span banks, credit unions, alternative lenders and fintechs and are addressed through a dedicated financial services line, with telecom operators, utilities, debt buyers, collection agencies and government sitting alongside as separate verticals rather than diluting it. Coverage across the lifecycle is equally wide, from pre delinquency through early and late stage collections to recovery, placement, litigation, bankruptcy and post charge off.