The application
The structured loan application and every revision of it, including who changed what and when.
Mortgage & lending
A loan origination system, usually shortened to LOS, is the software a lender runs a loan file on from application through to funding. It holds the file, tracks what stage it has reached, records every action taken on it, and connects out to the other systems the file depends on. When a lender says a loan is "in the system", the LOS is the system they mean.
An LOS is a system of record before it is anything else. Six things live inside it.
The structured loan application and every revision of it, including who changed what and when.
Everything collected against the file, tied to the condition or requirement it satisfies rather than sitting in a folder.
What underwriting has asked for, what has been provided, what has been cleared, and what remains open.
Which stage the file has reached, and how long it has been there. This is what makes a pipeline view possible.
Every action, timestamped and attributed. This is the part that matters when a regulator or an investor asks.
Borrower, loan officer, processor and underwriter, each with a role and a view appropriate to it.
Lenders name the stages differently, but the sequence is broadly consistent.
Application. The borrower's details are captured, usually through a point-of-sale system that hands off to the LOS.
Processing. Documents are requested and collected. Income, assets, employment and anything specific to the file. This stage is mostly waiting.
Underwriting. The file is evaluated, generally through an automated system first, and conditions are produced.
Condition clearing. Each condition is satisfied, verified and signed off individually. This is the longest stage in most files and the one that determines the timeline.
Closing. Final figures are issued, documents are prepared and signed.
Funding. The loan disburses, and the file leaves origination.
The LOS carries the file through all six. Its practical job is to make the answer to "where is this file and what is it waiting on" available without anyone having to ask.
An LOS is rarely the only system involved, and most confusion about what one does comes from mixing it up with what surrounds it.
In front of it sits the point-of-sale system, which is what a borrower actually interacts with. Application forms, document upload, status updates. The POS collects; the LOS holds.
Alongside it are the services the file depends on. Credit, verification, automated underwriting, pricing and product eligibility, document generation, compliance checking. Each is generally a separate provider, and the LOS calls out to them.
After it comes servicing, which takes over once the loan funds. That is a different system with a different job.
The reason this matters when evaluating an LOS: much of what a lender experiences as the LOS is actually the quality of these connections. A system that holds the file well but integrates poorly produces exactly the manual work it was bought to remove.
Origination automation is a broad term, and most of it falls into four categories.
Data movement. Getting information from one place to another without anyone retyping it. Application data into the LOS, extracted figures from documents, results back from a service that was called.
Rules and routing. Files assigned by criteria rather than by hand. Conditions generated automatically from findings. Tasks appearing for the right person at the right stage.
Status and chasing. Requests sent, reminders issued, and outstanding items tracked without someone maintaining a spreadsheet alongside the system.
Checks. Validations that run continuously rather than at review points, so a mismatch surfaces when it happens rather than three weeks later.
What is worth noticing is where automation has not reached. Document collection still depends on a borrower producing records. Verification still depends on a person reading a document and comparing it to a figure. Those two are where the calendar goes, and automating around them changes less than automating them would.
An account origination platform does the same job for deposit and banking products that an LOS does for loans.
A customer opening a checking account, a savings account or a business account goes through a process with the same shape: an application, identity checks, documents, a decision, and an account that exists at the end. The platform holds that file and moves it through.
The vocabulary differs because the products differ. There is no underwriting in the lending sense, identity and fraud checks carry more weight, and the timeline is compressed from weeks to minutes for a straightforward application.
Where the two converge is documents. Both are essentially document collection and verification workflows with a decision attached, which is why institutions running both often end up with the same problems in each: information keyed twice, records held in one system and needed in another.
Origination ends at funding. Everything after that is servicing, and it is a separate discipline.
A servicing system handles payments, escrow, statements, tax and insurance disbursement, delinquency, and the borrower relationship over the life of the loan. It runs for years where origination ran for weeks.
Loan servicing automation refers to that side: payment processing, escrow analysis, automated notices, self-service portals, and workflows around delinquency.
The handoff between the two is a known weak point. Documents collected during origination are frequently not available to servicing, so a borrower who verified something in March is asked for it again in October. The record exists; it just did not travel.
For an institution running both, the useful question is not which system is better but whether what one collected is visible to the other.
Four things get expected of an LOS that it does not deliver on its own.
It does not make credit decisions. Underwriting rules live in an automated underwriting system or with a human underwriter. The LOS records the outcome.
It does not verify documents. It stores them and tracks which condition they satisfy. Whether the figure on the pay stub matches the application is a person's job.
It does not fix a broken process. A workflow that was unclear on paper is unclear in software, and implementing an LOS around it usually makes the confusion more visible rather than less.
It does not remove data entry by itself. Unless information arrives structured or is extracted from the document it came in, someone is still typing it. This is the gap most lenders are surprised by after implementation.
The system is configured to match how work is done today, including the parts that were never deliberate.
Most of the value is in the connections, and most of the effort is too. A schedule built around the core system alone runs late.
The file lives in the LOS while the documents live in email and shared drives. Everything downstream inherits that split.
Old files that do not move cleanly become a second system nobody decommissions.
Processors keep their own spreadsheets when the system is slower for a task than the spreadsheet was. That habit is a signal, not a discipline problem.
An LOS knows a document exists and which condition it belongs to. It generally does not know what is inside it.
That distinction explains a surprising amount of what still feels manual after an implementation. A pay stub arrives, gets attached to the right condition, and the status updates. Somebody then opens it, reads the figure, compares it to the application, and records the result. The system tracked the document; a person did the work.
The consequences compound. Figures keyed by hand introduce mismatches that surface during verification. A borrower asked for the same document twice because two parts of the process cannot see each other. A condition cleared with no record of what satisfied it, which becomes an audit question later.
The layer that addresses this sits underneath rather than replacing anything: reading documents as they arrive, extracting the figures the file needs, and checking them against what was submitted. The LOS keeps its job. What changes is how much of the reading is done by a person.
Answers at a glance
A loan origination system, or LOS, is the software a lender runs a loan file on from application through to funding. It holds the file, tracks its stage, records every action, and connects out to credit, underwriting, document and compliance services.
LOS stands for loan origination system. In mortgage lending it refers to the platform holding the loan file from application to funding, as distinct from the point-of-sale system a borrower uses and the servicing system that takes over afterwards.
Application, processing, underwriting, condition clearing, closing and funding. Lenders name them differently, but the sequence is consistent. Condition clearing is usually the longest stage and the one that determines the timeline.
A point-of-sale system is what the borrower interacts with, handling application forms, document upload and status updates. The LOS is what the lender works in, holding the file and driving the workflow. The POS collects; the LOS holds.
Mainly data movement between systems, rules-based routing and task assignment, status tracking and chasing, and validation checks that run continuously. Document collection and document verification remain largely manual, and those are where most of the timeline goes.
An account origination platform does for deposit and banking products what an LOS does for loans. A customer opening an account goes through an application, identity checks, documents and a decision, and the platform holds that file and moves it through.
No. An LOS runs from application to funding. A servicing system takes over afterwards and handles payments, escrow, statements and the borrower relationship for the life of the loan. They are separate systems with separate jobs.
No. Credit decisions are made by an automated underwriting system or by a human underwriter. The LOS records the outcome, generates the resulting conditions and tracks them, but the decision itself happens elsewhere.
Usually because the system tracks documents without reading them. A document is attached to the right condition and the status updates, but a person still opens it, reads the figure and compares it to the application.
Integrations being underestimated, documents staying outside the system in email and shared drives, and the system being configured to match current process including the parts nobody chose deliberately.
An LOS knows a document exists and which condition it belongs to. It generally does not know what is inside it, which is why so much still runs on someone opening a file and reading a figure. CliQloan, one of the AmitaSoft platforms, works that layer: reading incoming documents, extracting the figures the file needs, and tracking what is still outstanding.
Read about CliQloan →