When a lender cannot reliably apply its own rules, every mortgage file is assigned by hand and every policy change waits on IT. This is how ML arteka rebuilt the decision.
“When a lender cannot apply its own rules automatically, the cost is not just slower decisions. It is inconsistent ones.”
A leading Canadian mortgage lender was assigning products, investors, and rates by hand. Compliance requirements, financial regulations, a broad product catalogue, and lender and brand rules made every underwriting decision complex, and a legacy system that could not reliably apply those adjudication rules pushed the work onto underwriters, adding time to every file and putting consistency and service levels at risk.
ML arteka replaced the legacy processing system with an intelligent decisioning approach that assigns the best mortgage loan from borrower information, lender policies, business rules, rates, discounts, and premiums, and expressed changing policies as business rules the lender’s own teams can manage without a system rebuild.
The result: 70% faster, more accurate decisions. An 87% reduction in underwriter manual effort. New investor onboarding cut by 60%, and business and product teams in control of their own rules.
Business Outcomes
These are business outcomes, measured in decision speed, underwriter capacity, and service-level attainment.
Shorter borrower wait, higher throughput per underwriter.
Capacity redeployed to complex, judgment-based cases.
Service commitments met substantially more often.
The Transformation
From a manual, rules-bound process to an intelligent decision the business can operate and evolve.
- 1
DiscoveryUnderstand the Decision, Not Just the SystemMapped how compliance rules, regulations, the product catalogue, and lender and brand rules actually shape an underwriting decision, and where the legacy system forced manual work. - 2
AutomationAutomate the DecisionIntroduced auto-decisioning to produce faster, more accurate mortgage decisions, assigning the best loan based on borrower information, lender policies, rates, discounts, and premiums. - 3
Business RulesMake Policy Changeable by the BusinessExpressed changing policies as business rules through an adjudication approach and optimized the decision flow, so a policy change no longer meant a system rebuild. - 4
EnablementHand Control to the BusinessDelivered an accessible interface so business and product teams manage investor, rate, product, and policy data directly, removing the dependency on IT for routine changes.
The Business Challenge
When Rules-Heavy Underwriting Outgrows the System
A leading Canadian mortgage lender could not consistently apply its own decisioning rules. Compliance, regulation, a large product catalogue, and lender and brand rules made underwriting complex, while manual assignment of products, investors, and rates slowed every file, strained service levels, and put decision consistency at risk.
Rules the system could not apply
The legacy system did not accurately factor adjudication rules across compliance, regulation, the product catalogue, and lender and brand rules.
Manual assignment on every file
Underwriters manually assigned products, investors, and applicable rates, adding significant time to every application.
Slow policy change
Changing a policy meant system work rather than a business action, slowing the lender’s ability to respond.
IT dependency for routine data
Business teams could not manage investors, rates, products, and policies without routing changes through IT.
Why Leadership Decided to Act
Looking to automate, leadership set out to improve underwriting and decisioning while enhancing the customer experience and strengthening market standing. The pressures driving the decision were clear:
- Manual assignment of products, investors, and rates was adding significant time to every file.
- A system that could not reliably apply adjudication rules put decision consistency and compliance at risk.
- Routine policy and data changes depended on IT, slowing the business down.
The ML arteka Approach
Model the Decision, Not Just the Workflow
- 1
FramingA Decision Problem, Not a Software ReplacementRather than digitizing the existing manual workflow, ML arteka modeled how policy, regulation, and lender rules produce a mortgage decision. - 2
RulesPolicy as Business-Managed RulesExpressed changing policies as rules the business can update without a system rebuild, keeping the decision current as policy evolves. - 3
CoverageFull Coverage of the Decision InputsThe best mortgage loan is assigned from borrower information, lender policies, business rules, rates, discounts, and premiums. - 4
Self-ServiceBusiness and Product Teams in ControlInvestor, rate, product, and policy management moved to the business with minimal outside dependency.
Business Outcomes in Detail
What the Numbers Mean
Faster, more accurate decisions shorten borrower wait times and increase the volume each underwriter can handle.
Removing manual product, investor, and rate assignment redeploys underwriter capacity toward complex cases that need human judgment.
Onboarding new lending investors faster expands the product and capital options available to borrowers.
Meeting service commitments more reliably improves both borrower experience and partner confidence.
Business and product teams manage investors, rates, products, and policies themselves, without routing routine changes through IT.
Technology Snapshot
Built Around the Mortgage Decision
The difference between modernization and transformation is who owns the rules afterward. By expressing mortgage policy as business-managed rules rather than embedding it in code, the lender gained a decision it can keep changing on its own. That is what turns a one-time project into a durable capability.
Executive Questions and Answers
The questions leadership tends to ask when evaluating an approach like this.
Speed
How can decisions be faster and more accurate at the same time?
When the system applies the rules directly instead of relying on underwriters to assign products, investors, and rates by hand, speed and accuracy improve together rather than trading off. In this engagement, decisions became 70% faster.
Compliance
What regulatory value does automated adjudication create?
Expressing policy as governed business rules produces a consistent, repeatable basis for each decision, which supports a defensible decision trail across compliance, regulation, and lender and brand rules.
Risk
What is the risk of leaving underwriting decisions to manual assignment?
Manual assignment of products, investors, and rates adds time to every file and introduces variation between underwriters. At the volume of a leading lender, that inconsistency becomes both a service-level risk and a compliance risk, because the same inputs can produce different decisions.
Implementation
How does an engagement like this unfold?
The work runs in phases: understanding the decision, automating it, expressing policy as business rules, and handing self-service control of rules and data to the business.
Scale
How does the solution scale as products and investors grow?
Because business and product teams manage rules and onboard investors themselves, the platform grows without a proportional increase in IT effort. The 60% reduction in new investor onboarding is one signal of that scalability.
