Equabli

The models are the differentiator, not the whole product. The company states it was built around predictive analytics from the ground up rather than adding intelligence to legacy software, and the product does not quite bear that out. Take the models away and a working product remains. It includes a recovery workflow platform and an orchestration portal for placing accounts with external agencies and law firms and tracking their performance.

A borrower self service payment channel and a document management system across the credit lifecycle complete it. Those are the components legacy collections software has always sold. What would be lost is the scoring layer that ranks accounts by repayment probability and net value, which is the differentiator and does rest on models. That is a platform with a model at its center rather than a model sold as a product.

The control is named, its scope is wide, and nothing about it is specified. Automated federal, state and local compliance checks are described as running across all collection activity. The scope extends beyond the platform's own actions to the activity of third party agencies and law firms working the accounts. The company positions this as giving a lender visibility into its recovery partners and reducing the need for external audits.

Oversight of other parties' conduct is a different proposition from guarding one's own. The platform also positions itself as highlighting risk and recommending the next best action rather than executing it, which implies the decision stays with the institution. No rule is named as checked, and there is no threshold, no escalation trigger and no description of what happens when a check fails.

Provenance is offered where validation should be. The published account of the models is that the founding team created them from prior industry experience, which explains their origin and says nothing about their behavior. No accuracy, lift, calibration or back testing evidence appears, and no retraining cadence or drift monitoring is described.

There is no ISO 42001 certification, and no validation documentation is offered to an institution defending a repayment probability model to its own examiner. The provenance claim also raises a question it never addresses: models built from experience gained at previous employers invite a question about what data informed them.

The most impressive number in the company's material does not belong to the company, and the distinction matters. Its predictive models are credited to a founding team who in previous roles generated more than $15 billion of collections across 100 million consumers. That is a legitimate provenance claim about where the modeling expertise came from, and it is not evidence of anything this platform has achieved.

No client is named, and no recovery uplift, liquidation rate or cost reduction figure is published for an actual deployment. A referenced customer story identifies no institution. The real signals are corporate: $6.35 million raised across two rounds, and more than fifty staff in five countries. An investment from a fund capitalized by community banks puts a party with money at stake behind the diligence.

No data boundary of any kind is described, and the platform's own architecture makes the question pointed. It aggregates delinquent portfolio data from multiple institutions and enriches it from outside sources, and it sits simultaneously between lenders, debt buyers and the agencies working accounts for both, so a single system holds competing parties' portfolio performance.

Nothing states whether one lender's recovery outcomes inform models scoring another's accounts, whether an agency's performance data is visible beyond the lender who placed with it, or whether a borrower appearing in two clients' portfolios is recognised across them. No red teaming, output screening or safety practice is described either.

The privacy surface is empty: no privacy policy detail, data processing agreement, subprocessor list or retention schedule is published. The data at stake is a bank's full delinquent account file enriched with additional sources, plus borrower documentation retained through the credit lifecycle. It covers people who are behind on payments and did not choose this vendor. Operating across five countries adds a cross border dimension that is never addressed. Nothing states retention periods for borrower documents, what enrichment sources are drawn on, or what happens to a lender's portfolio data when the relationship ends.

There is no public certification, attestation, trust center or security page, and what stands in their place is a statement that safeguarding client data is a top priority. That says what is published, not whether controls exist. The platform ingests a bank's internal loan management data, holds delinquent portfolio files and stores borrower documentation across the credit lifecycle. It publishes nothing about how any of it is protected, while its buyers are regulated institutions whose own vendor risk processes will ask.

The company is a software supplier to regulated lenders, agencies and debt buyers and holds no license itself, which is the correct posture. There is a tension in its stated value. That value includes automated compliance checking over third party recovery partners' conduct, positioned as reducing a bank's reliance on outside audits. The platform performs that oversight role without any supervisory standing, accreditation or independent assurance over its own checking. No regulator engagement, sandbox participation or examination is published.

No fairness testing, differential outcome analysis or model governance disclosure is published, against a scoring function whose consequences for a borrower are severe and largely invisible to them. The engine estimates repayment probability and net value per account and directs effort accordingly, which decides who is worked, who is pursued hardest, and whose account is routed to an external agency, a law firm or a debt sale.

Those are materially different outcomes for the person behind the account. Third party descriptions repeat a claim of equitable outcomes for borrowers and lenders, and nothing published tests it: no analysis of how scores or routing decisions distribute across demographic, geographic or product cohorts appears anywhere.

No liability position, error rate, remediation commitment or correction route is published. The structural position is the usual one for collections software: the platform scores the account and routes it, while the lender or agency remains the regulated party carrying the consequence of how the borrower is then treated.

Nothing states what happens when a score is wrong and an account that could have cured is routed to legal action or sold, and no borrower facing dispute or correction path exists. The compliance checking the platform performs is a control against a violation occurring, not a route for someone already harmed by one.

No model provider, family, version or hosting arrangement is disclosed. The material indicates the predictive models are proprietary and built in house, which is a partial answer about origin and the only one given. Nothing states whether any third party component sits in the scoring pipeline, where inference runs, or whether enrichment vendors supplying additional borrower data are part of the chain, which matters here because the platform explicitly enriches a lender's own records from outside sources it never identifies.

The integration story is specific about the right system and vague about which one. The platform is described as aggregating, standardizing and enriching data from a lender's internal sources including loan management systems, which is the correct system of record for this workload. Beyond that it maintains an integrated vendor network, routing placements to external agencies and law firms and pulling their activity back into one dashboard.

That makes it a two sided integration rather than a single feed. No loan management, core banking or agency platform is identified by name, no integration count is published, and there is no public API documentation.

The platform is described as cloud native and delivered as software as a service, and nothing further about deployment is published. No hosting regions, residency commitments, tenancy model or single tenant option appears. The company states operations across five countries serving a global client base, which makes residency a concrete question rather than a formality, and no statement addresses whether a European or Asian client's borrower data remains in region or where the scoring itself runs.

No pricing, tier structure or billing basis is published for any component of the suite. The gap is compounded by the modular structure, since the company markets multiple low commitment entry points that expand as a lender's delinquency grows, and a buyer cannot tell from published material whether the modules are separately licensed, how the price scales with portfolio size or placement volume, or what the expansion path actually costs. Nothing indicates whether the model is subscription, per account, or a share of amounts recovered, and those imply very different incentives.

Four distinct buyer types are addressed, and they sit on different sides of the same market. Banks and fintech lenders hold their own delinquency, debt buyers have purchased it, and collection agencies work it for both. Named verticals extend across online lending, auto finance and buy now pay later, as well as community banks, telecommunications and healthcare receivables. The company operates with more than fifty staff across five countries serving a global client base.

One of its investors is a fund capitalized by community banks, which is a meaningful signal about the buyer it is built for. No client is named anywhere, and no portfolio volume, placement count or market figure is published, so the breadth is described rather than evidenced.