DocVu.AI
DocVu.AI is Visionet Systems' document processing platform for mortgage lenders and servicers, launched in June 2021 and sold under its own brand. It reads incoming loan files before they reach the origination or servicing system. Its models classify mortgage documents, pull fields from pay stubs, tax returns, bank statements and closing packages, compare values across documents and flag what is missing. Each extracted field carries a confidence score, and low confidence values go to a person for verification before the lender's rules run and the data moves on.
Origination teams use it for borrower onboarding and income and asset checks, and quality teams for pre funding reviews and post closing audits. Servicers use it for bulk MSR transfers, collateral reconciliation, MERS and assignment tracking, loss mitigation files and foreclosure reviews. DMSVu adds cloud document storage with retention rules and intake through APIs, file transfer or Azure Data Share. A claims workflow applies the same engine to insurance eligibility checks and loss reporting.
Visionet reports more than 100 clients and 400 million pages processed a year. Its case studies describe a top five US bank cutting mortgage servicing costs by 53% in six months and a large servicer processing 200% more loans with 99.2% delivery accuracy, neither of them named. ServiceMac and Celebrity Financial are named users. Pricing is usage based and quoted, with no upfront cost.
Capability Axes
Capability grades
15 of 15 axes rated · 8 graded A or B
Classification, extraction and cross document checks are the work DocVu.AI is bought for. Machine learning identifies each document type, character recognition and language processing read values across pages and formats, and the engine compares those values between files and scores each one. Without the models the storage and workflow layers would have nothing to organize. A processor can pull the same income figure from a pay stub, a tax return and a bank statement together and see at once where they disagree.
Every extracted field carries a confidence score. Low confidence fields go to a human operator for verification while the rest continue, and cross document checks flag inconsistencies and missing items before a file reaches the loan system. Lender teams maintain the business rules and checklists that decide whether quality control on a file runs automatically or by hand, and an audit trail records each step. Credit and risk decisions, and how each flagged exception is resolved, stay with the lender's staff.
Each extracted value carries a confidence score, uncertain readings go to people for verification, and values are compared across documents before data moves on. Lender business rules run quality checks and exception handling, and audit trails let operations teams follow a file through review. Accuracy by document type, the model version in use and validation results have not been released.
Visionet's case studies report measured results for unnamed customers. A top five US bank cut its mortgage servicing costs by 53% within six months, and a large mortgage servicer processed 200% more loans in six months with 99.2% delivery accuracy while onboarding fell from months to weeks. Another institution saved more than 90% of the time spent on a manual post closing audit. ServiceMac credits Visionet with the quality of documents and data in loan boarding, and Celebrity Financial is a named implementation, both without figures.
Extraction combines models, lender business rules and human verification, and DMSVu lets an institution set how long stored files are kept and who can open them. Visionet describes the engine as tuned for mortgage work and improving continuously as it processes documents. Whether one lender's documents and corrections train models that serve other lenders is not stated, which leaves the cross client boundary open.
Loan files carry tax returns, bank statements, pay stubs and other nonpublic borrower information. DMSVu encrypts stored documents, controls access at cabinet, order and document level and applies retention rules the lender sets. The privacy notice on docvu.ai covers website visitors rather than borrower data processed for a lender, so subprocessors, deletion at the end of a contract and the lender's rights over borrower files sit in the services agreement.
DMSVu protects stored documents with encryption, access controls, multifactor authentication and disaster recovery, and keeps audit trails of file access and document events. Visionet refers to ISO 27001 and SOC controls across the company. A current certificate or attestation report that covers the extraction engine or DMSVu has not been named, and the insurance FAQ mentions only encryption and secure data handling.
Visionet sells document processing to regulated lenders and servicers and makes no credit or servicing decisions itself. Pre funding and post closing checks assemble evidence for the institution's own quality control program, where Visionet pitches the product at reducing repurchase demands and audit findings, and the audit trail shows how each file was handled. Decisions on applications and on resolving exceptions stay with the lender.
DocVu.AI's output feeds mortgage underwriting, so extraction errors that fell unevenly across borrower groups would carry into credit decisions. Visionet reports no fair lending testing, disparate impact analysis or accuracy breakdown by borrower group for the engine. The governance it describes covers file handling and data quality: confidence scores, human verification of uncertain fields, lender rules and audit trails.
A lender can challenge any extracted value inside its own workflow. Each field keeps its confidence score, uncertain readings go to an operator, a value can be checked against the file's other documents and corrected before it moves on, and the audit trail lets staff retrace a disputed value. No accuracy guarantee, reimbursement or indemnity for an extraction error stands behind the output.
Visionet describes the engine as character recognition, computer vision, language processing and machine learning tuned for mortgage documents, including recognition of handwriting and marks on forms, with rules and human review behind it. The site also mentions generative AI without naming a foundation model or an outside model provider. Azure Data Share is a route for exchanging files rather than a model supplier.
DocVu.AI feeds the systems of record lenders and servicers run. It sits upstream of ICE Mortgage Technology's Encompass for origination, and the servicing workflow maps verified data into MSP, the Black Knight servicing platform. MeridianLink lists Visionet as a partner for its LendingQB origination system.
DMSVu takes files in through APIs, secure file transfer and Azure Data Share and sends them back out through APIs and batch exports, and Alfresco and Salesforce are among the other named integrations.
DocVu.AI is delivered as a cloud service, and DMSVu stores documents in the cloud with several access tiers and controls at cabinet, order and document level. Visionet is headquartered in Cranbury, New Jersey, and runs a processing center in Pittsburgh, Pennsylvania. A hosting region for extraction and storage is not stated, and an on premises version is not described.
Visionet prices DocVu.AI by usage and quotes each deal, and the usage model starts with no upfront cost. Buyers can request pricing, run an ROI calculator, submit their volumes for a mortgage pricing calculation or start with a proof of concept at no cost. The rate per page, file or loan, any minimum volume and the charge for custom integration work are set in the quote.
Mortgage originators, lenders and servicers each have their own material, with separate solution pages for loan processing and for servicing. The servicing page walks through bulk MSR transfers, post transfer tracking, early default and loss mitigation files and foreclosure reviews, and quality teams use the same extraction for pre funding and post closing audits.
The insurance claims page is thinner, covering extraction, eligibility and valuation checks and duplicate or forgery checks without naming claim forms, systems or customers. Finance and accounting teams get an invoice workflow.
Alternatives to DocVu.AI
Other products that do the same kind of work as DocVu.AI, and what each one does.
Docsumo reads and checks documents for lenders, financial services companies and insurers. The New York company, founded in 2019, lets a lender send in an application with bank statements, pay stubs, tax returns and supporting files.
TRUE sells AI mortgage operations software to lenders, wholesalers and correspondents, mortgage insurers and loan buyers.
Blend sells digital origination software to banks, credit unions, independent mortgage banks and mortgage servicers. Borrowers apply through its app, and lenders work the file in tools behind it, across mortgage, home equity, consumer loans and deposit accounts.
AiCurio built what it describes as the first commercially available deep learning neural network for United States residential mortgage cash flows, predicting every monthly payment at the individual loan level including principal, interest and servicing costs, which necessarily means…
Fintor builds an agentic workforce for mortgage lenders and servicers, with agents that handle document collection, verification, compliance checks, closing coordination and post-close completeness across the loan lifecycle from intake onward.
Valon sells ValonOS, a system of record for mortgage servicing that brings the loan record, multi party accounting ledgers, money movement, workflows, reporting and compliance logic into one platform.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | Data Protection Terms | Implementation | Source |
|---|---|---|---|---|
|
Quoted. Usage based pricing with no upfront cost.
|
DocVu.AI is priced by usage under a quote from Visionet. The billable unit and the rate are set in the quote. | Data protection terms for borrower documents are part of the services agreement and are not tied to a pricing tier. | The usage model starts with no upfront cost, and Visionet offers a proof of concept at no charge. | Vendor Published |
Visionet quotes usage based pricing for DocVu.AI with no upfront cost. Buyers can request pricing, use an ROI calculator or submit their volumes for a mortgage pricing calculation, and a proof of concept is free. Rates for document storage, integration work and any minimum volume are set in the agreement.