Credit application referred for manual review

Plain language meaning

Credit application referred for manual review explains how a bank handles an AI-supported lending case when the score, affordability, KYC data, policy rules, fraud signals or explanation quality are not clean enough for automated approval or decline.

This topic is specifically about credit manual referral. It is not payment repair, AML investigation or generic case management.

This is the practical end of the AI and ML in Banking journey. The point is not to admire AI as a technology. The point is to understand how a bank uses AI inside real cases, with real customers, real queues, real controls, real risk owners and real evidence.

Where it sits in the banking AI journey

This card belongs to Practical AI and ML Scenarios. The working flow is Credit application, Score and policy check, Referral trigger, Underwriter review, and Decision evidence.

Read the flow as an operating story. Each stage has a system state, a data meaning, a control question, a responsible role, a possible exception, a customer or regulatory impact and a record that must survive audit. That is why the same AI idea looks very different inside a bank compared with a generic technology demo.

Banking data and evidence

The important data points are application ID, credit score, PD band, income evidence, affordability ratio, policy exception, reason code, and underwriter note. These items matter because they can change screening treatment, payment handling, credit decisions, investigation priority, policy interpretation, model response, career learning or operational closure.

The evidence pack should include application record, score output, policy checklist, manual review note, override reason, customer notice support, and decision trace. A strong bank can replay the case from source fact to AI support, deterministic rule, human action, final outcome and monitoring result. A weak bank only remembers that someone trusted a tool.

Controls that make AI adoption safe

The core controls are credit policy, affordability review, reason-code validation, fair-lending check, override authority, adverse-action control, and audit retention. These controls keep the chapter anchored to bank policy, customer protection, legal obligation, regulatory defensibility, model governance, operational resilience, privacy, security and auditability.

The design must define what AI may recommend, what it must not decide alone, where deterministic rules remain authoritative, who can approve or override, how evidence is retained, how errors are remediated and how learning is fed back safely.

Scenario and career lens

For practical scenarios, the learner should always ask what happened, what system detected it, what AI added, what policy or rule controlled the next step, who owned the decision, what customer impact existed and what record proves the final state.

For career topics, the learner should not reduce AI work to coding. Strong banking AI work also needs process mapping, data understanding, requirements clarity, controls thinking, testing skill, documentation discipline, regulatory awareness and the ability to explain consequences in plain language.

Regulatory and governance lens

Federal Reserve SR 26-2, dated 17 April 2026, gives revised model-risk guidance for traditional models and non-generative AI models used by banking organisations, including development, validation, monitoring, change control and governance.

The Federal Reserve's SR 26-3, dated 9 July 2026, highlights FinCEN's 12 June 2026 guidance on fraud-related information sharing under Section 314(b) for financial institutions subject to the BSA.

NIST AI RMF 1.0 uses Govern, Map, Measure and Manage functions for AI risk management, and NIST AI 600-1 adds generative-AI risk actions for grounding, privacy, cybersecurity, content provenance and human oversight.

BCBS 239 remains current for effective risk data aggregation and risk reporting, and the Basel Committee's January 2026 newsletter reiterates accurate, comprehensive and timely bank data capabilities.

The Basel Committee's operational resilience principles expect banks to identify, protect, respond, adapt, recover and learn when disruption affects critical operations.

U.S. Regulation B, 12 CFR 1002.9, requires specific principal reasons for adverse action in covered credit decisions, including when a creditor uses an AI model. CFPB Circular 2022-03 was withdrawn on 12 May 2025; do not cite it as current guidance. Primary sources: https://www.consumerfinance.gov/rules-policy/regulations/1002/9 and https://www.consumerfinance.gov/compliance/guidance/withdrawn-guidance/.

FFIEC BSA/AML examination guidance expects suspicious activity monitoring systems and independent testing to be risk-based, aligned to the bank's risk profile and supported by sufficient information for management and examiners.

OFAC's Framework for Compliance Commitments describes sanctions compliance programme components including management commitment, risk assessment, internal controls, testing and auditing, and training.

Diagram walkthrough

Read the diagram from left to right as Credit application, Score and policy check, Referral trigger, Underwriter review, and Decision evidence. It shows the practical route by which a case, role or learning step moves from input to controlled outcome.

Use it as a 30-minute study method. For each box, ask which system, data field, rule, owner, exception, customer impact and audit record belongs there. If the answer is unclear, that is the exact area to study again.

Most important mistake to avoid

The common failure is treating manual referral as a failure of automation. In credit, referral is often the correct control when the automated path cannot prove a fair, explainable and policy-aligned decision.

The correction is to stay narrow. Keep each scenario tied to its real banking process, keep every AI statement connected to evidence and keep the final answer useful for operations, risk, compliance, technology, product and learners.

A referral is a state, not an approval or denial

An applicant passes identity checks but has a short repayment history and conflicting income evidence. The credit service returns a referral under an approved policy. It records the score or PD version, feature snapshot, affordability findings, missing data, rule that caused referral and time. The customer can be told that more information or review is needed, subject to the product process. The bank should not present an unapproved credit limit or a confident adverse reason before the decision is made.

An underwriter checks the original documents, account credits, existing obligations and any bureau information that was available at application time. They may request evidence, correct a source error, approve within delegated authority or decline under policy. If a source is corrected, recalculate affected features and preserve the first calculation for audit. An override records the original model output, reason, evidence, approver and final decision. Repeated overrides on the same thin-file segment may show that the model or policy is poorly calibrated, but an override rate by itself does not establish that conclusion.

Test a late bureau response, income in another currency, a loan exactly at a policy threshold and a reviewer whose approval authority has expired. Confirm that no booking or disbursement occurs while the referral remains open. For covered U.S. credit decisions, an adverse action requires specific principal reasons under Regulation B; applicability elsewhere needs its own assessment. Reconstruct what the underwriter saw, which rule was decisive, what the customer received, and how the final action reached the core lending and reporting systems.

The referral queue should have a service owner and an aging rule. A customer who supplies the requested evidence must not be scored indefinitely on the stale first snapshot. Conversely, a later account credit should not be backfilled into an explanation of the initial referral. The final case record should show both moments and the rule in force at each one. Monitor whether particular customer groups are referred more often because of missing bureau coverage or inconsistent data feeds, and investigate the cause before changing a threshold. This makes a human review substantive rather than a rubber stamp.

Banking practice note: banking purpose

For credit application referred for manual review, banking purpose must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from application ID to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

AI can classify, rank, compare, retrieve, summarise, suggest, warn and help a human work faster. It should not invent facts, replace sanctions disposition, weaken AML judgment, bypass fraud authority, change payment data without approval, decide credit outcomes without explainability or create career confidence without real banking understanding.

A strong implementation records the source event, data fields, model or prompt version, rule result, score or generated output, threshold band, user action, override reason, customer communication, monitoring signal and closure evidence. That record lets a bank explain the case without relying on memory.

Banking practice note: source system

For credit application referred for manual review, source system must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from credit score to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: data field meaning

For credit application referred for manual review, data field meaning must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from PD band to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: AI support boundary

For credit application referred for manual review, AI support boundary must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from income evidence to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: deterministic rule

For credit application referred for manual review, deterministic rule must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from affordability ratio to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: human authority

For credit application referred for manual review, human authority must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from policy exception to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: customer impact

For credit application referred for manual review, customer impact must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from reason code to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: regulatory impact

For credit application referred for manual review, regulatory impact must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from underwriter note to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: privacy and security

For credit application referred for manual review, privacy and security must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from application ID to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: audit replay

For credit application referred for manual review, audit replay must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from credit score to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: exception handling

For credit application referred for manual review, exception handling must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from PD band to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: false-positive control

For credit application referred for manual review, false-positive control must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from income evidence to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: false-negative control

For credit application referred for manual review, false-negative control must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from affordability ratio to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: screening separation

For credit application referred for manual review, screening separation must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from policy exception to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: fraud separation

For credit application referred for manual review, fraud separation must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from reason code to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: AML separation

For credit application referred for manual review, AML separation must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from underwriter note to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: payment operation

For credit application referred for manual review, payment operation must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from application ID to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: credit policy

For credit application referred for manual review, credit policy must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from credit score to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: policy source

For credit application referred for manual review, policy source must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from PD band to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: model version

For credit application referred for manual review, model version must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from income evidence to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: prompt version

For credit application referred for manual review, prompt version must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from affordability ratio to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: drift monitoring

For credit application referred for manual review, drift monitoring must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from policy exception to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: root cause

For credit application referred for manual review, root cause must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from reason code to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: quality sampling

For credit application referred for manual review, quality sampling must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from underwriter note to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: role accountability

For credit application referred for manual review, role accountability must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from application ID to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: learning output

For credit application referred for manual review, learning output must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from credit score to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: test scenario

For credit application referred for manual review, test scenario must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from PD band to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: communication quality

For credit application referred for manual review, communication quality must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from income evidence to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: fallback path

For credit application referred for manual review, fallback path must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from affordability ratio to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: closure evidence

For credit application referred for manual review, closure evidence must be treated as a practical banking concern. It decides whether the AI support is connected to the right process, the right owner, the right data and the right customer or regulatory outcome.

Trace one item from policy exception to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: banking purpose

Trace one item from reason code to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: source system

Trace one item from underwriter note to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: data field meaning

Trace one item from application ID to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: AI support boundary

Trace one item from credit score to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: deterministic rule

Trace one item from PD band to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: human authority

Trace one item from income evidence to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: customer impact

Trace one item from affordability ratio to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: regulatory impact

Trace one item from policy exception to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: privacy and security

Trace one item from reason code to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: audit replay

Trace one item from underwriter note to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: exception handling

Trace one item from application ID to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: false-positive control

Trace one item from credit score to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: false-negative control

Trace one item from PD band to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: screening separation

Trace one item from income evidence to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: fraud separation

Trace one item from affordability ratio to policy checklist. Then ask which control from reason-code validation proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: AML separation

Trace one item from policy exception to manual review note. Then ask which control from fair-lending check proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: payment operation

Trace one item from reason code to override reason. Then ask which control from override authority proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: credit policy

Trace one item from underwriter note to customer notice support. Then ask which control from adverse-action control proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: policy source

Trace one item from application ID to decision trace. Then ask which control from audit retention proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: model version

Trace one item from credit score to application record. Then ask which control from credit policy proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.

Banking practice note: prompt version

Trace one item from PD band to score output. Then ask which control from affordability review proves the item was valid, timely, authorised, relevant and retained. If that trace cannot be shown, the scenario is not ready for production or serious study.