Ocrolus

Reading a thousand document types at stated accuracy above 99 percent regardless of format or quality, then deriving income, cash flow and fraud signals from them, is model work with no rules based alternative, and the measured effect is the giveaway: a named lender cut bank statement review from five hours to about ten minutes. Apply the removal test and what remains is manual document review, which is the incumbent process being displaced. One newer component, the conditioning engine, is described as deterministic by design, which is a deliberate exception rather than the pattern.

The positioning is explicit and comes from a customer rather than the marketing team, with a named executive stating that the platform is not meant to replace underwriters but to increase their output and reduce mistakes, and the company's own published commentary argues for finding the balance between automation and human expertise rather than maximising automation.

An underwriting dashboard supports review, and discrepancy detection surfaces conflicts between borrower documents and application data for a person to resolve. Less described is the newer automated conditioning capability, which generates underwriting conditions and matches supporting documents without a stated approval step.

Three external signals give a validator more than most vendors provide. Accuracy above 99 percent is published as a figure. The government sponsored enterprise approval means the income analysis has been examined by a party with money at stake. And insuring data capture accuracy through the Lloyd's market means an underwriter independently priced the error rate, which is a form of third party validation almost nobody in this index can point to. Deterministic conditioning is inspectable by construction. Still absent are the methodology behind the accuracy figure, error rates by document type, model documentation and a validation summary.

The strongest evidence surface in this lane. Volume is stated at roughly 750,000 credit applications a month, customers are named across every tier from large fintech lenders and payment companies to community banks and credit unions, and outcomes are attributed to named institutions with named executives on the record: bank statement review down more than 90 percent and time to close down 23 percent at one lender, 8,500 staff hours and 90,000 dollars of processing expense saved annually at a community bank, processing time down about 90 percent at another, underwriter time in file down 29 percent at a third. One caution belongs on the record: the published customer count moved from more than 500 in late 2024 to more than 400 in early 2026, a decrease worth understanding rather than ignoring.

One design decision deserves specific credit. The conditioning engine is described as deterministic by design and grounded in selling guide requirements rather than generative, which is a deliberate choice to keep a component that maps directly onto agency rules out of probabilistic territory, and very few vendors in this index explain where they chose not to use a model. Accuracy above 99 percent is published rather than asserted vaguely. What is not addressed is the training boundary: nothing states whether borrower documents processed for one lender inform models serving another, and no model providers are identified.

The documents this platform reads are the most revealing financial records a person holds. A bank statement discloses where someone shops, worships, seeks medical care and sends money, and Ocrolus processes them at a rate of roughly 750,000 applications a month alongside pay stubs and tax forms.

No published privacy framework, retention schedule, subprocessor list or statement of financial privacy service provider obligations was located, which is a conspicuous omission at that volume and sensitivity.

No trust centre, enumerated certification list, attestation scope or audit period was located in this pass. Customers include large payment companies, national lenders, community banks and credit unions, all of which run vendor assurance programmes that would require attestations before borrower financial documents moved, so the actual control environment is certainly stronger than the published record shows.

A top grade earned through a specific formal approval rather than through general compliance language. Income analysis produced by the platform is approved by the government sponsored mortgage enterprise and its outputs are eligible for representations and warranties relief, which is a defined programme with published criteria and real legal consequence: it shifts repurchase risk away from the lender for income calculation defects.

Passing that assessment means a party other than the vendor and the customer has examined the analysis and staked something on it. Product scope also maps onto agency selling guide requirements and federal loan programme rules directly.

The inclusion case is real and stated by a customer rather than the vendor, with a community bank executive describing an expanded lending base reaching borrowers with non traditional income sources, which is exactly what cash flow underwriting is supposed to achieve.

The unexamined side is that bank statement analysis surfaces spending patterns, transaction categories and deposit sources that correlate with neighbourhood, employment sector and immigration status, and extraction accuracy itself varies with document quality, which is worse for gig and self employed borrowers than for salaried ones. Lending sits under equal credit opportunity rules, and no fair lending testing, demographic accuracy breakdown or adverse action documentation was located.

The most substantive liability position found anywhere in this index, and it rests on two independent risk transfers rather than a promise. Data capture accuracy is insured through the Lloyd's market, meaning a third party underwriter assessed the error rate and stands behind it financially, which is stronger than a self issued guarantee because someone with no interest in the marketing priced the risk.

Separately, income outputs eligible for representations and warranties relief move repurchase exposure away from the lender for income calculation defects. Both protect the institution rather than the borrower: a wrongly declined applicant still has no route to see or contest the extracted figures behind the decision.

The data chain is short and clean by nature, since inputs are documents the borrower supplies and records already held in the lender's origination system rather than purchased third party data, which removes a whole class of fourth party exposure other vendors in this index carry. The model layer is undisclosed. No providers are named for document classification, extraction or fraud detection, no subprocessor list is published, and nothing states where borrower documents are processed once uploaded.

Integration runs into the system of record rather than beside it. Direct integration with the dominant mortgage loan origination system covers document sync, indexing and matching, so output lands where underwriters already work and automated conditions attach to the live loan file, which is the mortgage equivalent of core banking integration and the hard part of this market.

Public interface documentation supports direct build, and a lending software partner embeds the technology for its own customer base, extending reach to institutions that would not procure it directly.

Delivery is cloud hosted software as a service serving domestic lenders, so cross border complexity is limited compared with the global vendors in this index. Residency still matters given the material: borrower bank statements, pay stubs and tax forms are retained within loan files subject to record keeping requirements measured in years. No hosting regions, tenancy separation, residency options or subprocessor chain were located in this pass.

No rates, tiers, billing unit or minimum were located. The unit question is straightforward here and still unanswered, since a platform processing 750,000 applications monthly is almost certainly priced per document or per application with volume banding, and the published customer outcomes are all denominated in hours and dollars saved with no cost placed against them.

Coverage spans lending in its full breadth, addressing mortgage, small business funding, consumer lending and auto finance with distinct material for each, and the named customer list runs from large technology lenders and payment companies through mortgage originators to community banks and agricultural credit institutions, which is unusual reach across institution size. Adjacent verticals including legal, healthcare benefits, tax and tenant screening extend it further without diluting the lending focus.