See the file. Understand the relationships.
Move the decision forward.
Sentinel brings together deterministic underwriting for non-qualified mortgage (Non-QM) and business-purpose lending, business intelligence, and mortgage ontology. Its three platforms help lending teams evaluate loan files, investigate relationships, and document decisions.
Explore three complementary platforms.
Each supports a distinct part of the lending process.
Sentinel Origination is a deterministic automated underwriting system for Non-QM and business-purpose real estate lenders. Current workflows focus on debt service coverage ratio (DSCR) lending, with planned support for rehabilitation (rehab), ground-up construction, and bridge loans. It applies version-controlled lender rules to verified loan inputs and produces traceable findings, calculations, and conditions for underwriter review.
Sentinel Origination is an automated underwriting system (AUS) for Non-QM and business-purpose real estate lenders. Current debt service coverage ratio (DSCR) workflows connect loan data, document evidence, and lender guidelines. Planned support for rehabilitation (rehab), ground-up construction, and bridge loans will extend the AUS with program-specific rules and document requirements.
Sentinel evaluates verified loan inputs against a version-controlled set of lender guidelines and deterministic rules. With the same inputs, engine version, guideline version, and evaluation date, the engine produces the same rule outcomes.
Findings identify the applicable rules, supporting calculations, missing inputs, and required conditions. OCR-assisted extraction supplies evidence for review; it does not replace the underwriting rules. The lender retains final decision authority.
Lender requirements that can be evaluated automatically are encoded across five implementation layers. Automated tests check alignment with source guidelines, behavior at policy thresholds, and expected rule outcomes. Regression tests help identify unintended changes.
A structured guideline library records policy sections and the published guideline version.
Executable rules evaluate verified inputs and return explicit findings or unresolved requirements.
Program thresholds, eligibility matrices, and applicable overlays refine the evaluation.
Evidence requirements, document validity rules, and condition mappings define the work still required.
Each recorded run retains its inputs, engine version, guideline version, and findings for traceability.
For lenders that update policy regularly, planned version selection will allow a loan to be evaluated against the current guideline package or a retained prior package. Historical replay will use the original inputs, evaluation date, and compatible engine version; a new evaluation under revised guidelines will be recorded as a separate run.
Retain the guideline edition, applicable evaluation date, inputs, and findings associated with each run.
A newer guideline edition does not rewrite an earlier result. Each executable edition must be retained and validated before it can be selected.
Currently, Sentinel evaluates the guideline edition configured in the engine and retains the version associated with each historical run. Evaluating other editions requires the corresponding executable rule packages.
Tests verify encoded behavior and help detect unintended changes. Requirements that call for human judgment remain part of the lender’s review.
Optical character recognition (OCR) and document analysis extract text, dates, and key loan data from uploaded files. Extracted evidence supports document classification, cross-document reconciliation, and reviewer validation.
EXTRACT / CLASSIFY / RECONCILEEvaluate document age and expiration against applicable guideline requirements and the underwriting reference date. Distinguish age limits from stated validity dates, such as insurance coverage periods and ID expiration.
DATED / CURRENT / EXPIRED / UNRESOLVEDReview the loan package for required documents, pages, fields, signatures, and other evidence of document execution. Flag missing information, inconsistent values, and document-integrity concerns for correction or underwriter review.
EVIDENCE REQUIRED / REVIEW REQUIREDAn anonymized AUS report identified a shortfall under a market-rent stress scenario, a financing discrepancy, an invalid lease date, and concerns about an editable electronic form. It distinguished unresolved rule evaluations from conditions requiring additional documentation or underwriting judgment.
The output includes priority findings, underwriting calculations, unresolved rule evaluations, a conditions register with owners and milestones, a document inventory, and an audit trail tied to the guideline of record.
Sentinel supports connectivity with ICE Encompass and Salesforce to bring lending workflows and relevant records together through configured integrations.
Sentinel analyzes loan documents and background records to explain each finding, identify supporting evidence, and highlight unresolved questions.
Review principal identities, aliases, sanctions and exclusions, litigation, liens, and background records, with the scope and status of each check documented.
A / PRINCIPALS & GUARANTORSMap direct and indirect ownership, related businesses, trusts, shared addresses, and transaction counterparties to clarify entity relationships and control.
B–C / ENTITIES & RELATIONSHIPSReview outstanding conditions, severity ratings, supporting evidence, an analytical summary, and revision history. Missing evidence remains visible.
S / CONDITIONS & SYNTHESISA vendor fraud score was elevated. The investigation separated neighborhood distress signals from evidence about the actual transaction.
CONTEXT CLARIFIED · REVIEW STILL REQUIREDMarket distress, inconsistencies in reported rental experience, and a recently formed borrowing entity called for closer review.
Title records indicated that the property had been held for an extended period and supported seller independence. Valuation reviews were consistent. An operating business network helped explain the principal’s background.
Complete entity background and exclusions checks, reconcile title vesting, and document how the vendor alerts were resolved. Favorable evidence did not erase unresolved conditions.
A result from one report revision, not a performance benchmark.
Incomplete records raised questions about entity formation, beneficial ownership, and possible equity extraction.
Formation filings established an operating history. Executed ownership documents clarified control. A draw ledger and bank records explained rehabilitation funding and contradicted the initial concern about cash returned to the borrower.
A property-type conflict persisted between the origination record and appraisal. The report retained it as a high-priority correction, alongside outstanding identity and documentation items.
Findings supported by sources. Unresolved questions made explicit. A record updated as evidence changes.Examples are adapted from Sentinel Business Intelligence reports. Client identities and transaction identifiers have been omitted. Business Intelligence supports review; the lender retains responsibility for the credit decision.
Request a Business Intelligence demoA loan is more than a collection of fields. It is a network of people, assets, obligations, and evidence.
Bring the application and its source documents into view.
Examine ownership, counterparties, and public-record signals.
See the relationships between the loan and its evidence.
Give reviewers the context for their next decision.
SOURCE DOCUMENTS → FILE CONTEXT → RELATIONSHIP CONTEXT → INFORMED REVIEW
Built for the people accountable
for what happens next.
BI findings reference documents and public records so reviewers can trace the evidence supporting a conclusion.
The ontology connects entities, guarantors, collateral, and conditions to reveal how the case fits together.
Intelligence supports review. Origination keeps underwriting and the decision record within the lending workflow.