StandardC

ApplyC, VerifyC and MonitorC shipped as a workflow suite built on GPS tagged virtual site visits, document capture and screening before StandardC AI was layered on top. Remove the agents and a saleable site inspection, onboarding and monitoring system remains.

Agents run compliance checks and the initial review of an application or loan file against the institution's policies, a structured report goes to examiners, and human reviewers disposition the findings. Thresholds, escalation routes and what happens when a reviewer misses an agent error are not published.

Consistent, structured reports show reproducibility, which is not correctness: an agent can be reproducibly wrong. No accuracy figure, test result or validation package is published for banks expected to validate vendor models.

Delene Gilbert, Vice President at Prism Bank, is quoted on VerifyC, while the other quotes come from the firm's own chairman and chief experience officer. Claims of 20 times the capacity and 80 to 90 percent savings on virtual site visits are attributed to no institution and carry no method.

Customer personally identifiable information never reaches an AI model, per a patent pending privacy architecture, which settles the training question by construction. Whether screening results or merchant intelligence from one institution inform what another receives is not stated.

Compliance with the Gramm Leach Bliley Act and 256 bit AES encryption are stated, and the trust center lists retention, disposal and data classification policies available on request. Retention schedule and subprocessor list are unpublished.

Secureframe hosts a trust center showing a completed SOC 2 Type 2 audit, with the report, penetration test results and continuity plans released under nondisclosure on request. ISO 27001 is mentioned without any claim that the firm holds it.

A provider doing onboarding and monitoring for supervised banks falls within interagency third party risk expectations, and whether the firm has faced a bank service provider examination is not stated. Vendor due diligence documents, from an insurance certificate to a state registration, are available on request.

Loan files in ApplyC's origination flow are reviewed by agents, which places the product inside equal credit and fair lending duties whenever a review shapes a credit outcome. Fairness testing, bias review and adverse action content are unpublished, and name screening adds uneven error rates across scripts.

Small businesses in cannabis related or high risk merchant segments, with few other banks willing to serve them, may have files reviewed first by an agent. Examiners can trace the decision, but the applicant sees no sign of automated review and has no route to contest it, and no warranty or remedy is published.

Model, provider, version and hosting are all unnamed, and the trust center lists no subprocessors. Testing the privacy architecture claim requires knowing which model runs where, under whose terms.

ID.me identity verification is recorded into the business's audit trail, the one named integration. Not one core banking platform is named, though the workflow has to reach the core where an account opens.

Deployment in days is published, but no hosting provider, cloud region or tenancy model is named. A product built on where data goes says nothing about where its own platform runs.

Pay per use and enterprise plans are named, but no unit, rate or minimum, so the shape of the contract is visible and the cost is not. The usage unit could be applications, site visits or agent reports.

Community banks, credit unions and merchant processors are the named buyers, with ApplyC for origination, VerifyC for virtual site visits and MonitorC for screening. Coverage is United States only, and no large bank, asset manager or insurer appears.