Infrrd

Extraction without templates is the product, and it cannot exist without inference. Infrrd uses deep learning, computer vision and language processing, with the proprietary Titan platform and machine learning character recognition. Strip the models out and the platform has nothing to operate on, unlike a workflow platform that keeps working. Financial services is not an industry page but a set of separately named, separately purchasable products with domain rules encoded. The company runs a general purpose business alongside them.

Two halves pull in opposite directions. The strong half is a control that cites the rule it enforces. Tolerance tests under the federal TRID disclosure rule, and comparisons between the closing disclosure and the loan estimate, run at the point of upload. Tamper resistant logs and version control are kept for regulators and investors. The weak half is the disposal path. Exceptions are routed to a person and, in the vendor's own words, everything else flows straight to delivery.

No confidence threshold, review rate or escalation criterion is published. For an extraction product feeding an underwriting file, the question is not whether exceptions get seen but what decides that a field is not an exception, and that is exactly what is missing.

A public self serve demo invites a prospect to upload a difficult document and see what the system reads. That lets a buyer generate their own accuracy evidence rather than take a figure on trust. Alongside it sit validation against business rules, audit trails, version control and tamper resistant logs. No extraction accuracy rate is published. The content marketing establishes the human baseline, citing manual processing error rates of one to two percent.

The product page then claims better than human accuracy without ever publishing its own rate. Naming the comparison point and withholding your own side of it is a more sophisticated version of the unquantified accuracy claim. It borrows the credibility of a number the vendor never has to stand behind.

State National is the one named customer with a described scope and a quote, an insurance program business whose intake spans forms from 2,100 separate carriers. A partner is quoted on the MortgageCheckai loan quality control launch. Gartner and Everest Group each name Infrrd a Leader in their 2026 evaluations, from a field of eighteen vendors in one case. Analyst placement shows market recognition rather than measured performance.

The quantified claims that do exist (approvals sixty percent faster, loan review twice as fast) are attached to no institution. A customer stories library exists on the site and its contents are not reflected here.

Nothing is published on whether customer documents train models. The question has real force because the platform explicitly allows an organization to create and train new models. The documents are loan files and claim files containing borrower financial detail and medical reports. No statement isolates one customer's training data from another's or describes retention after extraction. Redaction is offered as a customer controlled feature, which is a control over output rather than a commitment about the vendor's own use.

The privacy posture is more specific than the usual gesture. The vendor names the principles it operates under and enumerates the individual rights it supports: access, erasure, portability and objection. It names standard contractual clauses as its transfer mechanism and ships redaction as a control. That is materially more than a line saying the product was built with privacy in mind.

The gap is specific and important: there is no position at all on the Gramm Leach Bliley Act (GLBA), even though the mortgage product extracts borrower income, assets and liabilities, the textbook definition of nonpublic personal information.

Infrrd has a SOC 2 report and names the three trust service criteria it covers explicitly rather than leaving them implied. Continuous control monitoring and annual audits support it. A standing security page describes encryption in transit and at rest, fine grained access management, isolation and redaction. That is a real posture rather than a badge row. The report is available on request rather than through a trust portal, and no bridge letter covers the gap between audit periods. No ISO 27001 certification is held, and no penetration testing is disclosed.

Unlicensed and unsupervised, the norm for a document processing vendor. It operates close to named regulatory machinery, encoding federal mortgage disclosure tolerance tests and building files to be examinable by regulators and investors. The lender remains the party answerable to regulators.

No fairness position, no governance disclosure. The exposure is not abstract: extracted income, asset and liability values feed underwriting decisions on individual borrowers, and a separate capability flags fraud patterns and discrepancies on loan files. An automated fraud pattern flag is a consequential inference about a named person, and nothing describes how it is calibrated or who reviews it.

No recourse route and no allocation of responsibility. The specific chain matters here: a misread income figure passes through automated validation, is not classified as an exception, and flows to underwriting and then to investor delivery. The audit trail is thorough and records what was extracted rather than whether it was correct, so the very completeness of the log can make a wrong value look verified.

Proprietary and patented, with the model classes named at the broadest level and no model, provider, architecture or version identified anywhere. The patent position stands in for disclosure as the differentiator. A patent proves originality to a lawyer and tells a model risk function nothing about what is running.

Integration is described as the real work rather than glossed, and the vendor says so itself. It writes that mapping extracted fields into a loan origination system, a core banking platform or a compliance database is consistently underestimated and frequently becomes the primary implementation cost. Exception queues, investor delivery and a partnership with a mortgage compliance specialist are all named. No specific system, connector or protocol is named, so the depth is asserted by category rather than demonstrated by counterparty.

Residency is addressed rather than ignored. Regional hosting options are offered and standard contractual clauses are named for transfers. The security material describes tenant isolation, hardened cloud infrastructure and customer controlled policies. No region is named, and no single tenant or customer hosted option is described. The hosting options are offered rather than specified.

No price, tier or unit is published. One commitment sits next to the absence: the service level is presented as a customer choice on a stated range from fifteen minutes to twenty four hours. That is a real operational term, published openly. A buyer can learn the turnaround they will get and nothing about what it costs.

Within financial services, buyers are mortgage lenders, servicers and investors, quality control and compliance teams, and insurance carriers. The work spans origination, quality control, post close and servicing, plus claims and policy administration. Volume evidence is real: more than a thousand document types, twenty two languages and more than a hundred billion pages. The institution types are narrower than the page count suggests. No banking, capital markets or wealth buyer is addressed, and a material share of the business sits outside financial services, in invoice processing, construction and manufacturing.