Biz2X

Take out the learned components and a working digital lending platform remains, which is what several named banks bought before those components existed. The core is loan origination and servicing software. It handles application intake, document collection and workflow, along with configurable decision rules, scorecards and portfolio servicing. All of that is deterministic, and it was the whole product at launch in 2019.

The learned components sit on top: AI enabled underwriting decision support and risk scoring, a machine learning borrower health report, and AI that fills SBA forms and keeps them current with program rules.

Control sits with the institution by design: decision rules and user permissions are configurable, scorecards are built to the buyer's own credit policy, applications are banker assisted, and the decision engine can be limited to instant suggestions for the underwriting team rather than automated decisions. Every lending decision is logged automatically with the data behind it.

One part of the flow runs the other way: predefined rejection rules automatically screen out applicants who fail basic criteria before a person sees them, and Biz2X describes no review of those automated declines, no threshold and no exception path. An automated knockout is the point in a lending workflow where oversight matters most.

Transparency is claimed as a product benefit and no model documentation supports it. AI enabled underwriting decision support and a machine learning borrower health report are described with no model named, and no validation method, performance measurement, monitoring, versioning or drift disclosure appears for any component.

What the institution does control is the scorecard and rule layer, which is configured to its own credit policy, so the deterministic half is governable by the buyer while the learned half is undocumented. The published figures are throughput claims about approval speed rather than evidence about model quality.

A named institution carries a quantified outcome: UMB Bank's commercial lending program increased loan processing rates and cut decision time to 15 days after integrating the decision engine with its core provider. HSBC is named in two case studies, one for automated decisions on United States small business applications of $100,000 or less and one for Canadian small businesses, with a five minute application and automated approvals up to C$250,000.

Popular Bank runs fully automated lending up to $100,000 and uses the Biz2X scoring model to underwrite smaller deals. Biz2X's own headline claim is that it cuts SBA approval times by 50 percent, a figure no case study isolates.

The platform grew out of the system Biz2Credit built for its own online business lending, and Biz2X publishes no separation between client institutions' application, financial and outcome data and that lending operation, and no position on training data, retention for model development or third party model providers. On the platform side, borrower data sits in storage that is versioned and encrypted, with continuous backups and point in time restoration.

Biz2X sets out no retention, deletion, subprocessor or cross border transfer terms for a platform that collects borrower financial records and bank transaction data. The white label arrangement adds a question the material never addresses: the borrower experiences the institution's brand throughout and may not know a third party is processing the application at all.

Biz2X says it is SOC 2 and ISO 27001 certified and has passed rigorous audits, without naming the report type, the auditor or the certification body or posting either document. Its AWS environment runs Security Hub, GuardDuty, Inspector, Macie and Detective, with AWS Config and Systems Manager, and logging is enabled across the stack.

A software vendor holding no license or supervised standing. It operates close to a government lending program, building eligibility checks, standard operating procedure compliance, form filling and direct submission into the administration's system, but conforming a product to a program's rules is not authorization under them, and the participating lender remains the approved party throughout.

Biz2X puts automation inside small business credit decisions: scorecards set to each lender's credit policies, a rules engine that can run entirely automated lending decisions, AI enabled underwriting decision support and risk scoring, and predefined rejection rules that knock out applicants who miss basic criteria before a person sees them. Nothing on disparate impact, or on how adverse action reasons are produced for those knockouts, is published. Every lending decision is logged automatically with the data behind it, which gives a lender the decision record a fair lending review would draw on.

A small business screened out by an automated rejection rule deals with the lender whose brand it sees, and Biz2X publishes no route for that applicant to an explanation or a review and no allocation of responsibility between platform and lender. The decision trail stays with the lender, recorded with the data behind each decision at every stage, so an institution can reconstruct a knockout it configured.

AI and machine learning are referenced throughout with no model, provider, family or version named, and no statement of which capabilities use which. AWS is named as the cloud whose security services the platform relies on, which is an infrastructure relationship rather than a disclosure of what performs the inference.

One integration is named in a customer result: UMB Bank's decision engine runs integrated with its incumbent core provider. The platform is built on a microservices architecture so it can connect across a wider technology estate. The core banking platforms themselves go unnamed, so a buyer cannot check whether its own core is supported without asking. It does submit directly into the United States Small Business Administration's E-Tran system, a named external system of record rather than a category.

A turnkey software as a service platform on AWS, Biz2X is built on microservices, with data stored in each client's own geography and a configurable white label environment so the institution's brand fronts the borrower experience. A new client can be running in 90 to 120 days, and the same timeline applies to changes after go live.

Biz2X publishes no prices, tiers, per application or per loan rates, no implementation cost indication and no distinction in commercial terms between the core platform, the government program line and the AI features. Every route ends in a demo or a consultation.

Banks, credit unions and other lenders of all sizes in the United States and beyond buy Biz2X, with HSBC in the United States and Canada, Popular Bank and UMB Bank among its named users and Tata Capital using it for small business lending in India; no customer count or portfolio volume is published.

The focus is small and midsize business lending, with a dedicated line for United States Small Business Administration programs that need program specific eligibility and compliance handling, so the platform goes deep in one lending category rather than wide across many, on a design built from more than a decade of Biz2Credit's own online lending.