100 Best Gemini Credit Risk Analysis Prompts for Banks

Direct Answer

The best Gemini prompts for credit risk analysis for banks are designed to systematically identify, quantify, and mitigate borrower and portfolio risk using structured outputs, auditable processes, and clear actions. They specify data sources, desired formats, and constraints to ensure repeatability and governance, while avoiding ambiguity in risk interpretation. These prompts help staff generate consistent analyses and evidence-based decisions.

Introduction

This page provides a practical, topic-specific prompt library for credit risk analysis in banks. It helps risk teams, data scientists, and decision-makers perform rigorous, repeatable assessments.

The prompts cover borrower-level signals, portfolio-level risk, macro factors, model governance, and remediation actions, all tailored to banking contexts.

How to Use These Gemini Prompts

• Replace placeholders like [data], [segment], and [scenarios] with your internal datasets and targets.
• Add constraints or formats you require, such as a 1-page summary, a CSV table, or a bullet-point risk plan.
• Request outputs to include explicit recommended actions and owners to ensure accountability.
• Verify outputs by cross-checking against historical performance and scenario testing.
• If needed, combine multiple prompts to build a multi-scenario risk assessment.
• Regularly review prompts for data quality and regulatory changes.

100 Best Gemini Credit Risk Analysis Prompts for Banks

1. Borrower credit trend: You are a risk analyst. Task: Analyze borrower credit trend for a given customer segment. Context: Use last 24 months of repayment histories, utilization, and payment delinquency data from [data]. Output: structured brief including trend direction, drivers, and recommended actions. Constraints: exclude non-bank sourced data, limit to 400 words, provide a one-sentence risk rating.
2. Portfolio default hotspot: You are a credit risk advisor. Task: Identify default hotspot across the bank's SME loan portfolio. Context: Use portfolio-level exposures, sector distribution, and historical default rates from [data]. Output: a ranked list of top 3 hotspots with causes, indicators, and recommended mitigations. Constraints: deliver in bullet list, max 250 words.
3. Liquidity-adjusted beta: You are a risk modeler. Task: Estimate liquidity-adjusted beta for the retail loan book. Context: Use [data] including liquidity metrics, cash flow stability, and repayment timing. Output: a methodology brief and final beta value with confidence interval. Constraints: include formula, document assumptions.
4. Credit utilization jump: You are a credit risk analyst. Task: Detect sharp increases in credit utilization prior to delinquency. Context: Use [data] from merchants and card portfolios. Output: a summary of signals, thresholds, and recommended monitoring actions. Constraints: provide a 5-point signal checklist.
5. Payment behavior patterns: You are a risk insights researcher. Task: Map payment behavior patterns linked to future default. Context: Use [data], segment by product and segment. Output: observed patterns, correlation scores, and action plan. Constraints: avoid overfitting, report limitations.
6. Covenant risk check: You are a covenant monitoring specialist. Task: Assess covenant breach risk for new loan approvals. Context: Use [data] including covenants, liquidity coverage, and financial covenants. Output: risk rating, breach likelihood, recommended covenant adjustments. Constraints: provide tabular output.
7. LGD driver map: You are a loss given default analyst. Task: Map drivers of LGD for secured vs unsecured loans. Context: Use [data] including collateral values, cure rates, and recovery timelines. Output: comparative driver map with numeric weights. Constraints: present in a compact matrix.
8. Stress-test scenarios: You are a stress testing facilitator. Task: Run macroeconomic shock scenarios for credit risk. Context: Use [data] and a baseline. Output: scenario descriptions, impact estimates on default and loss, and mitigation notes. Constraints: provide at least two scenarios.
9. Grading model audit: You are a model validator. Task: Audit the credit scoring model for bias and stability. Context: Use [data], model version, and performance metrics. Output: audit findings, bias checks, and remediation plan. Constraints: include recommended monitoring.
10. Automated alerts: You are a risk operations specialist. Task: Design automated alerts for rising risk indicators. Context: Use [data] including delinquency signals and exposure. Output: alert rules, thresholds, and escalation paths. Constraints: ensure explainability of alerts.
11. PD calibration: You are a calibration engineer. Task: Calibrate probability of default for new portfolio. Context: Use [data] and historical PD curves. Output: calibration method, adjusted PDs, and backtesting plan. Constraints: show goodness-of-fit metrics.
12. Exposure concentration: You are a risk analyst. Task: Quantify exposure concentration risk by industry and geography. Context: Use [data] including current exposures. Output: concentration scores, top 5 concentrations, and mitigation steps. Constraints: include exposure limits.
13. Credit policy gap: You are a policy analyst. Task: Identify gaps between current credit policy and observed risk. Context: Use [data] and regulatory constraints. Output: a prioritized gap list with recommended policy updates. Constraints: provide impact estimates.
14. Alt-data signals: You are a data scientist. Task: Integrate alternative data signals for credit risk. Context: Use [data] sources such as payment histories and utilities. Output: validation results, added predictive value, and caveats. Constraints: document data quality.
15. Riskt score explainability: You are an interpretability specialist. Task: Produce plain-language explanations for risk scores. Context: Use [data], model outputs, and segmentation. Output: customer-friendly rationale and audit notes. Constraints: avoid jargon.
16. Remediation plan: You are a remediation advisor. Task: Propose remediation plans for at-risk borrowers. Context: Use [data], borrower status, and repayment options. Output: prioritized actions with expected impact and cost. Constraints: provide time horizons.
17. Counterparty risk: You are a risk analyst. Task: Evaluate counterparty credit risk for financing partners. Context: Use [data] including exposure and creditworthiness. Output: risk rating, exposure limits, and monitoring plan. Constraints: include escalation triggers.
18. Credit migration: You are a migration analyst. Task: Track credit rating migrations and their impact. Context: Use [data] including historical ratings. Output: migration matrix and risk-adjusted expected loss. Constraints: report trends.
19. Recovery curve fit: You are a recovery analyst. Task: Fit recovery curves for different collateral types. Context: Use [data] with historical recoveries. Output: curves, key drivers, and interpretation notes. Constraints: compare collateral types.
20. Account-level risk: You are an account risk auditor. Task: Assess risk at the borrower account level. Context: Use [data] including account history and product mix. Output: risk score per account and recommended actions. Constraints: maintain confidentiality.
21. Remittance risk: You are a risk researcher. Task: Analyze remittance patterns for repayment risk. Context: Use [data] including cash flows and geography. Output: risk flags and policy suggestions. Constraints: ensure data privacy.
22. Fraud-redit blend: You are a fraud-risk analyst. Task: Separate fraud indicators from credit risk signals. Context: Use [data] including transaction patterns. Output: joint risk assessment with fraud flags and credit risk scores. Constraints: avoid false alarms.
23. Cumulative risk view: You are a risk strategist. Task: Build a cumulative risk view across products. Context: Use [data] including exposure and delinquencies. Output: dashboard-ready summary with warning levels. Constraints: keep to 1 page.
24. Vendor credit check: You are a vendor risk analyst. Task: Assess credit risk of external vendors. Context: Use [data] including payment history and credit terms. Output: risk rating and monitoring plan. Constraints: include dependency notes.
25. UBS-style risk memo: You are a risk writer. Task: Produce a concise risk memo for senior management. Context: Use [data], portfolio metrics, and scenarios. Output: executive summary with key risks and actions. Constraints: two-page maximum.
26. Early delinquency predictor: You are a data scientist. Task: Build an early delinquency predictor. Context: Use [data] including lead indicators. Output: model description, features, and performance. Constraints: validate on holdout sample.
27. Credit uplift potential: You are a portfolio manager. Task: Identify uplift opportunities from credit line reviews. Context: Use [data] including utilisation and repayment timelines. Output: recommended actions and expected impact. Constraints: quantify impact.
28. Risk-adjusted pricing: You are a pricing analyst. Task: Propose risk-adjusted pricing for new loans. Context: Use [data] including PD and LGD. Output: pricing table and rationale. Constraints: align with policy.
29. Stress-ready provisioning: You are a provisioning specialist. Task: Estimate provisions under stress. Context: Use [data], regulatory buffers, and historical losses. Output: provision figures and methodology. Constraints: show sensitivity.
30. Credit cycle timing: You are a macro risk analyst. Task: Predict credit cycle timing for lending segments. Context: Use macro indicators and internal data. Output: timeline, risk flags, and mitigations. Constraints: provide confidence intervals.
31. Geography risk map: You are a risk mapper. Task: Map geographic risk differences for credit. Context: Use [data] with location data. Output: choropleth-style risk map description and top regions. Constraints: avoid PII.
32. Product-level risk: You are a product analyst. Task: Compare risk across loan products. Context: Use [data] by product. Output: risk scores, default rates, and suggested product tweaks. Constraints: limit to five products.
33. Collateral value drift: You are a collateral analyst. Task: Monitor collateral value drift over time. Context: Use [data] including appraisals and market data. Output: drift report with triggers for reevaluation. Constraints: provide action thresholds.
34. Regulatory-ready report: You are a regulatory liaison. Task: Prepare data pack for regulatory reporting. Context: Use [data], templates, and controls. Output: structured report with narratives and metrics. Constraints: ensure traceability.
35. Credit line utilization: You are a credit risk officer. Task: Analyze line utilization efficiency. Context: Use [data] including drawdown timings and limits. Output: utilization score and optimization plan. Constraints: propose limit revisions.
36. Shopfloor-risk review: You are a risk reviewer. Task: Conduct a quick shopfloor risk review of a loan desk. Context: Use [data] including processes and controls. Output: findings, gaps, and quick fixes. Constraints: one-page report.
37. SME sector stress: You are a macro-analyst. Task: Stress SME sector defaults under interest-rate shocks. Context: Use [data] and sector data. Output: default projections and hedge suggestions. Constraints: scenario-bound.
38. Credit quality scorecard: You are a scorecard designer. Task: Design a credit quality scorecard for new applicants. Context: Use [data] including features and outcomes. Output: scorecard scheme, features, and validation plan. Constraints: maintain fairness checks.
39. Treasury spillover: You are a risk economist. Task: Assess liquidity spillovers from credit events. Context: Use [data] and funding costs. Output: spillover impact report and mitigation steps. Constraints: quantify liquidity impact.
40. Recovery lag analysis: You are a recovery analyst. Task: Analyze recovery lags by collateral type. Context: Use [data] with historical recoveries. Output: lag estimates, confidence, and policy notes. Constraints: compare unsecured.
41. New credit policy test: You are a policy tester. Task: Run what-if tests on a revised policy. Context: Use [data], scenarios, and controls. Output: risk impact and compliance notes. Constraints: provide pass/fail criteria.
42. Historical default drill: You are a risk historian. Task: Reconstruct historical defaults and drivers. Context: Use [data] archive. Output: key drivers, timing, and lessons. Constraints: anonymize data.
43. Portfolio heatmap: You are a portfolio analyst. Task: Produce a risk heatmap for the loan book. Context: Use [data] with risk scores and exposures. Output: heatmap description and top alerts. Constraints: one-page summary.
44. Qualitative risk notes: You are a credit auditor. Task: Add qualitative risk notes for high-risk borrowers. Context: Use [data] and reviewer judgments. Output: notes with justification and next steps. Constraints: maintain objectivity.
45. Countercyclical buffer: You are a risk strategist. Task: Propose countercyclical buffer adjustments. Context: Use [data], macro indicators, and exposures. Output: buffer levels, rationale, and monitoring. Constraints: link to policy.
46. Bank-health proxy: You are a risk analyst. Task: Build a bank-health proxy using credit metrics. Context: Use [data] and peer benchmarks. Output: score, interpretation, and actions. Constraints: ensure comparability.
47. Delinquency delta: You are a data analyst. Task: Compute delinquency delta across cohorts. Context: Use [data] with cohort definitions. Output: delta values and insights. Constraints: present clearly.
48. Credit line renewal: You are a renewal risk specialist. Task: Assess renewal risk for expiring lines. Context: Use [data] including history and affordability. Output: renewal risk rating and recommended terms. Constraints: avoid bias.
49. Fraud-flag filter: You are a fraud analyst. Task: Build a credit risk filter that flags fraud signals in credit applications. Context: Use [data], include validation plan. Output: rules, thresholds, and review steps. Constraints: minimize false positives.
50. LGD sensitivity: You are a risk modeler. Task: Sensitize LGD to collateral haircuts. Context: Use [data] and haircut assumptions. Output: sensitivity table and recommendations. Constraints: show ranges.
51. Cash-flow stress: You are a cash-flow analyst. Task: Stress cash flows under scenarios for borrower liquidity. Context: Use [data]. Output: impact on debt-service coverage ratio and actions. Constraints: present in a table.
52. Credit-score drift: You are a data scientist. Task: Track credit-score drift over time by cohort. Context: Use [data]. Output: drift graph description and triggers. Constraints: provide numeric thresholds.
53. New-to-bank risk: You are a risk analyst. Task: Assess risk for new-to-bank customers. Context: Use [data] including onboarding signals. Output: risk rating and onboarding controls. Constraints: ensure regulatory compliance.
54. Prospect due-diligence: You are a diligence analyst. Task: Conduct due-diligence prompts for large borrowers. Context: Use [data], financials, and governance. Output: diligence checklist and risk rating. Constraints: avoid bias.
55. Collateral sufficiency: You are a collateral analyst. Task: Validate collateral sufficiency for secured lending. Context: Use [data] including valuations and insurance. Output: sufficiency verdict and actions. Constraints: document valuation method.
56. Remediation cost estimate: You are a cost estimator. Task: Estimate remediation costs for at-risk loans. Context: Use [data] including modification options. Output: costed plan and impact on NPV. Constraints: provide confidence intervals.
57. Portfolio aging: You are a portfolio manager. Task: Analyze aging of credit portfolio. Context: Use [data] including performance by vintage. Output: aging chart description and actions. Constraints: concise.
58. Recovery-value forecast: You are a recovery forecaster. Task: Forecast recovery values under varying conditions. Context: Use [data] including past recoveries. Output: forecast table and risk notes. Constraints: scenario-based.
59. Regulatory mapping: You are a compliance analyst. Task: Map risk metrics to regulatory requirements. Context: Use [data], dashboards, and controls. Output: compliance checklist and gaps. Constraints: stay current.
60. Credit risk dashboard: You are a dashboard designer. Task: Build a credit risk dashboard blueprint. Context: Use [data] including metrics and alerts. Output: widgets, data sources, and refresh cadence. Constraints: be actionable.
61. Early-stage screening: You are a screening analyst. Task: Create early-stage credit screening prompts. Context: Use [data] and onboarding signals. Output: screening criteria and escalation paths. Constraints: fast decisioning.
62. Cross-border risk: You are a risk officer. Task: Assess cross-border lending risk. Context: Use [data] including currency and regulatory differences. Output: risk rating and hedging plan. Constraints: comply with local laws.
63. Consumer-prospect test: You are a consumer risk analyst. Task: Test risk signals on consumer loan prospects. Context: Use [data], compare segments. Output: signal set and expected default rates. Constraints: avoid overclaiming.
64. Vendor-sovereign risk: You are a risk analyst. Task: Evaluate external vendor and sovereign risk interactions. Context: Use [data] including exposure by country. Output: composite risk score and monitoring steps. Constraints: frame limitations.
65. Recovery timing: You are a recovery strategist. Task: Estimate timing of recoveries for different collateral. Context: Use [data] with histories. Output: timing bands and actions. Constraints: provide confidence.
66. Credit-writing prompt: You are a credit writer. Task: Draft clear risk notes for a loan file. Context: Use [data], assessment and rationale. Output: executive summary and risk flags. Constraints: concise.
67. Behavioral signals: You are a behavioral analytics analyst. Task: Identify behavioral signals predicting default. Context: Use [data] including usage patterns. Output: signal list and action plan. Constraints: avoid bias.
68. Cycle-adjusted PD: You are a modeler. Task: Compute cycle-adjusted PD for portfolio. Context: Use [data], historical cycles. Output: adjusted PD values and justification. Constraints: compare to baseline.
69. Pricing-approval pack: You are a pricing pack creator. Task: Assemble risk-informed pricing justification. Context: Use [data], PD, LGD. Output: package with tables and narrative. Constraints: include sensitivity.
70. Delinquency cascade: You are a risk analyst. Task: Model delinquency cascades across stages. Context: Use [data] and stage definitions. Output: cascade chart and mitigation steps. Constraints: keep to one page.
71. Sector-default risk: You are a sector risk expert. Task: Compare default risk across sectors. Context: Use [data] with historical defaults. Output: ranking and reasons. Constraints: avoid overgeneralization.
72. Credit-life cycle: You are a lifecycle analyst. Task: Map credit risk across lifecycle stages. Context: Use [data] including product journey. Output: risk at each stage and controls. Constraints: minimize handoffs.
73. Credit-quality alerts: You are a risk operator. Task: Define quality-alert thresholds for new applications. Context: Use [data] including features. Output: alert matrix and action steps. Constraints: deterministic thresholds.
74. Model-agnostic checks: You are a model validator. Task: Propose checks that work across models. Context: Use [data] and performance metrics. Output: a checklist and reporting template. Constraints: generalize.
75. New-portfolio onboarding: You are a onboarding risk analyst. Task: Assess risk before onboarding new portfolio. Context: Use [data], client information, and onboarding controls. Output: risk rating and required verifications. Constraints: minimize onboarding friction.
76. Credit line tiering: You are a policy designer. Task: Propose tiered credit lines by risk. Context: Use [data] and governance. Output: tier structure, terms, and approvals. Constraints: ensure fairness.
77. Sanctions screening: You are a compliance risk analyst. Task: Screen credit applicants for sanctions risk. Context: Use [data] and external feeds. Output: sanctions risk verdict and records. Constraints: avoid false positives.
78. Recovery outcome audit: You are a recovery auditor. Task: Audit recovery outcomes by loan type. Context: Use [data] and recovery payments. Output: audit findings and improvement plan. Constraints: maintain confidentiality.
79. CLTV estimation: You are a credit strategist. Task: Estimate customer lifetime value for loans. Context: Use [data], segment data, and churn. Output: CLTV calculations and strategy. Constraints: include assumptions.
80. Early warning index: You are a risk analyst. Task: Build early warning index from signals. Context: Use [data] across features. Output: index formula and thresholds. Constraints: interpretable.
81. Capital adequacy tie: You are a risk economist. Task: Tie credit risk to capital adequacy metrics. Context: Use [data] and regulatory math. Output: risk-adjusted capital estimates and notes. Constraints: transparent methodology.
82. Credit policy simulator: You are a policy simulator. Task: Simulate policy changes and risk outcomes. Context: Use [data], scenarios, and controls. Output: impact matrix and recommended actions. Constraints: provide break-even points.
83. Resilience scoring: You are a resilience modeler. Task: Score borrower resilience to shocks. Context: Use [data], income volatility, and liquidity. Output: resilience score and plan. Constraints: avoid overreliance on one metric.
84. Portfolio diversification: You are a diversification analyst. Task: Recommend diversification to reduce risk. Context: Use [data] including exposures. Output: diversification plan and expected risk reduction. Constraints: quantify effects.
85. Cost-to-income effect: You are a financial analyst. Task: Assess cost-to-income impact of risk mitigations. Context: Use [data], operating costs. Output: ROI and payback period. Constraints: conservative estimates.
86. Waterfall of losses: You are a losses analyst. Task: Model loss waterfalls under multiple defaults. Context: Use [data], recovery, and timing. Output: waterfall diagram description and numbers. Constraints: keep simple.
87. Fraud-resilient underwriting: You are a risk underwriter. Task: Build underwriting prompts resilient to fraud. Context: Use [data] including validation steps. Output: underwriting rules and checks. Constraints: minimize fraud.
88. Productivity prompts: You are a risk ops lead. Task: Streamline risk reporting prompts for teams. Context: Use [data], existing processes. Output: standardized prompts and templates. Constraints: ensure reproducibility.
89. Credit-risk glossary: You are a knowledge engineer. Task: Generate a glossary of credit risk terms for internal use. Context: Use [data] and standard definitions. Output: term definitions and examples. Constraints: concise.
90. Audit trail prompts: You are a governance specialist. Task: Create prompts that generate audit trails for risk decisions. Context: Use [data], decisions, and authors. Output: auditable narrative and timestamps. Constraints:immutable log.
91. Ethical risk prompts: You are an ethics-focused risk analyst. Task: Inspect prompts for fairness and bias. Context: Use [data] including demographic features. Output: bias findings and mitigations. Constraints: comply with laws.
92. Forecast reconciliation: You are a reconciliation analyst. Task: Reconcile forecasted defaults with actuals. Context: Use [data], time series. Output: reconciliation report and corrective steps. Constraints: explain deviations.
93. Credit event calendar: You are a risk planner. Task: Create a calendar of credit events to monitor. Context: Use [data] and regulatory dates. Output: event list with owners and triggers. Constraints: keep current.
94. Audit-ready dashboards: You are a dashboard engineer. Task: Build dashboards suitable for audits. Context: Use [data], governance controls. Output: dashboard specs and data lineage. Constraints: exportable.
95. Scenario-combine prompts: You are a scenario designer. Task: Combine multiple scenarios into a composite analysis. Context: Use [data], three to five scenarios. Output: composite risk score and narrative. Constraints: avoid overcomplication.
96. Data-quality checks: You are a data steward. Task: Define data-quality prompts for credit risk data. Context: Use [data] quality metrics. Output: checks, tolerances, and remediation steps. Constraints: monitor continuously.
97. Explainability brief: You are an explainability writer. Task: Produce plain-language explainability notes for a risk model. Context: Use [data] and outputs. Output: user-friendly explanation and risk flags. Constraints: avoid jargon.
98. Regulatory narrative: You are a regulatory writer. Task: Draft concise regulatory narratives for lending portfolios. Context: Use [data], metrics, and controls. Output: narrative with metrics and controls. Constraints: align with guidelines.
99. Audit-ready file pack: You are an risk documentation specialist. Task: Assemble a complete audit-ready file pack for a loan. Context: Use [data], decisions, and evidence. Output: organized pack with sections and references. Constraints: ensure traceability.
100. Credit-risk playbook: You are a risk playbook author. Task: Compile a practical playbook for credit risk management. Context: Use [data], governance, and workflow. Output: step-by-step playbook with roles and controls. Constraints: keep actionable.

Best Practices

• Keep prompts modular and chainable so you can reuse components across analyses.
• Explicitly demand explainability sections for every risk score and decision.
• Require auditable outputs with data references and versioning for compliance.
• Test prompts with small cohorts before applying to full portfolios.
• Annotate outputs with caveats and known uncertainties to avoid over-interpretation.

Common Mistakes to Avoid

• Assuming model outputs are perfect without validating with historical outcomes.
• Using vague prompts that lack data sources or required outputs.
• Overloading prompts with too many hypotheses, reducing clarity.
• Ignoring data quality and governance requirements in risk prompts.
• Failing to specify action owners and timelines for recommended mitigations.

FAQ

What is the goal of these Gemini prompts for banks?

To deliver repeatable, auditable credit risk analyses that identify borrower and portfolio risk, support decisioning, and align with regulatory expectations.

How should I use placeholders in prompts?

Replace placeholders with your internal data, customer segments, products, and scenarios to tailor the outputs to your bank.

What outputs should each prompt produce?

A clear risk assessment, supporting metrics, and actionable recommendations with owners, time horizons, and any required formats.

How can I ensure outputs are explainable?

Request plain-language rationales, feature importance summaries, and visual aids where appropriate, plus notes on limitations.

How do I validate prompt results?

Backtest against historical outcomes, compare to existing risk reports, and run scenario analyses to check robustness.

Can prompts be used for regulatory reporting?

Yes, with structured outputs, data lineage, and narrative disclosures that align to regulatory expectations.