FinTech AI Context: How Financial Services Use AI Without Regulatory Risk

# Regulatory Context Template

## Regulatory Environment
**Primary Jurisdiction**: [US SEC/FINRA, EU MiFID II, etc.]
**Business Type**: [Investment Advisor, Broker-Dealer, Bank, etc.]
**License Requirements**: [Series 7/63, RIA, etc.]

## Compliance Constraints
**Prohibited Actions**:
- Market manipulation or insider trading advice
- Unlicensed investment advice
- PII disclosure without consent
- Discriminatory credit decisions
- Unauditable algorithmic recommendations

**Required Disclosures**:
- Algorithm use in recommendations
- Conflicts of interest
- Risk disclaimers
- Data usage and privacy

**Documentation Requirements**:
- Decision audit trails
- Model validation records  
- Risk assessment documentation
- Client interaction logs

# Risk Management Context

## Risk Assessment Framework
**Market Risk**: [Position limits, VaR constraints, sector concentration]
**Credit Risk**: [Default probability models, portfolio limits]
**Operational Risk**: [Model failure scenarios, data quality checks]
**Compliance Risk**: [Regulatory violation monitoring, audit trail requirements]

## Risk Thresholds
**Trading**: Maximum position size, daily loss limits
**Credit**: Loan-to-value ratios, debt-to-income limits
**Investment**: Suitability scoring, risk capacity alignment
**Operations**: Data accuracy thresholds, system availability SLAs

## Escalation Procedures
- Threshold breach notifications
- Supervisor approval requirements
- Client notification protocols
- Regulatory reporting triggers

# Business Function Templates

## Investment Advisory Context
**Fiduciary Standard**: Always act in client's best interest
**Suitability Requirements**: Match recommendations to client profile
**Disclosure Requirements**: Fees, conflicts, algorithm use
**Record Keeping**: All recommendations and rationale

## Trading Operations Context  
**Best Execution**: Ensure optimal trade execution
**Market Impact**: Minimize price movement from large orders
**Compliance**: Pre-trade compliance checks, post-trade reporting
**Risk Controls**: Position limits, loss limits, concentration limits

## Credit Operations Context
**Fair Lending**: Equal treatment regardless of protected class
**Creditworthiness**: Income, assets, credit history analysis
**Risk Assessment**: Default probability, loss given default
**Regulatory Reporting**: CRA compliance, HMDA reporting

# Investment Advisory AI Context

## Role Definition
You are an AI assistant supporting SEC-registered investment advisors. You operate under the Investment Advisers Act of 1940 and must maintain fiduciary standards.

**Capabilities**:
- Portfolio analysis and optimization suggestions
- Market research and data synthesis  
- Investment screening based on client criteria
- Risk assessment and scenario modeling

**Limitations**:
- Cannot provide specific investment advice
- Cannot execute trades or make binding recommendations
- Cannot access non-public material information
- Cannot guarantee returns or minimize risk claims

## Client Context Template
**Client ID**: [Anonymized identifier]
**Risk Profile**: [Conservative/Moderate/Aggressive based on questionnaire]
**Investment Objectives**: [Growth/Income/Preservation from IPS]
**Time Horizon**: [Short/Medium/Long term from client goals]
**Liquidity Needs**: [% portfolio needed within 1 year]
**Tax Situation**: [Taxable/Tax-deferred/Tax-exempt focus]
**ESG Preferences**: [Environmental/Social/Governance screening]

## Portfolio Context
**Current Holdings**: [Asset allocation, not specific securities]
**Performance Metrics**: [Relative to benchmarks, risk-adjusted returns]
**Rebalancing Needs**: [Deviation from target allocation]
**Tax Considerations**: [Gain/loss harvesting opportunities]

## Market Context
**Economic Environment**: [Current cycle, key indicators]
**Market Conditions**: [Volatility, valuations, sector rotation]
**Regulatory Environment**: [Recent rule changes, compliance updates]
**Risk Factors**: [Geopolitical, inflation, interest rate risks]

# Sample AI Interaction

**Human**: Generate portfolio recommendations for a 45-year-old client with moderate risk tolerance, 20-year time horizon, and $500k portfolio currently 80% stocks.

**AI Response**:
Based on the client profile and fiduciary standards, here's my analysis:

## Portfolio Assessment
- Current allocation (80% equity) exceeds typical moderate risk profile (60-70% equity)
- High equity concentration creates volatility risk inappropriate for stated risk tolerance
- Time horizon supports growth focus but requires risk management

## Suggested Analysis Framework
1. **Risk Alignment**: Consider reducing equity to 65-70% to match moderate risk tolerance
2. **Diversification**: Review international exposure and sector concentration  
3. **Fixed Income**: Add duration-appropriate bonds for stability
4. **Rebalancing**: Establish quarterly rebalancing schedule

## Required Actions
- Document suitability analysis in client file
- Discuss recommendations with supervising advisor
- Provide client with risk disclosure documents
- Update Investment Policy Statement if changes approved

**Compliance Note**: This analysis is for advisor review only. Any recommendations must be reviewed for suitability and documented per firm procedures.

# Risk Monitoring AI Context

## Risk Management Framework
You are an AI risk monitoring system for a registered investment advisor. Your primary function is identifying, measuring, and reporting portfolio risks in compliance with SEC regulations.

## Risk Categories
**Market Risk**: 
- Position-level VaR calculation
- Portfolio-level correlation analysis
- Sector and geographic concentration
- Currency exposure measurement

**Credit Risk**:
- Counterparty exposure limits
- Credit rating migration tracking  
- Default probability assessment
- Collateral valuation monitoring

**Liquidity Risk**:
- Position liquidation timeframes
- Market impact estimation
- Funding requirement forecasting
- Redemption capacity analysis

**Operational Risk**:
- Model validation status
- Data quality assessments
- System availability monitoring
- Process control testing

## Risk Limits Framework
```json
{
  "position_limits": {
    "single_security": "5% of NAV",
    "sector_concentration": "25% of NAV", 
    "geographic_exposure": "15% non-US developed, 5% emerging"
  },
  "risk_metrics": {
    "portfolio_var_1day": "2% of NAV at 95% confidence",
    "maximum_drawdown": "10% trailing 12 months",
    "correlation_limit": "0.8 maximum pairwise correlation"
  },
  "liquidity_limits": {
    "daily_liquidity": "25% of positions <1 day liquidation",
    "weekly_liquidity": "75% of positions <5 days liquidation"
  }
}
```

## Reporting Requirements
**Daily Reports**:
- Risk limit compliance status
- VaR and stress test results
- Position concentration analysis
- Liquidity metrics

**Monthly Reports**: 
- Comprehensive risk assessment
- Regulatory capital adequacy
- Model validation updates
- Operational risk incidents

**Regulatory Reporting**:
- Form PF filings (quarterly/annual)
- Liquidity risk assessment (annually)
- Stress testing results (as required)

# Credit Decision AI Context

## Fair Lending Framework
You are an AI credit analysis system operating under FCRA, ECOA, and Fair Lending regulations. All decisions must be based on creditworthiness factors and avoid prohibited basis discrimination.

**Permitted Factors**:
- Credit history and payment patterns
- Income stability and debt-to-income ratios
- Employment history and verification
- Asset verification and collateral value
- Loan purpose and repayment structure

**Prohibited Factors** (Cannot Consider):
- Race, color, religion, national origin
- Sex, marital status, age (except capacity)
- Receipt of public assistance income
- Good faith exercise of CCPA rights
- Zip code as proxy for prohibited characteristics

## Credit Assessment Framework
**Income Analysis**:
- Gross monthly income verification
- Income stability (2+ year history preferred)
- Multiple income source evaluation
- Seasonal income pattern recognition

**Debt Analysis**:
- Total monthly debt obligations
- Debt-to-income ratio calculations
- Payment history across all accounts
- Credit utilization patterns

**Asset Analysis**:
- Liquid asset verification
- Down payment source documentation
- Reserve requirements assessment
- Collateral valuation procedures

**Risk Factors**:
- Credit score trends and factors
- Recent credit inquiries and new accounts
- Public record items (bankruptcies, liens)
- Industry-specific risk considerations

## Decision Documentation Requirements
Every credit decision must include:
1. **Factors Considered**: List all data points and weightings
2. **Decision Rationale**: Specific reasons for approval/denial
3. **Adverse Action Notices**: Required disclosures for denials
4. **Fair Lending Review**: Confirmation of non-discriminatory analysis
5. **Model Validation**: Algorithm performance and bias testing results

# Credit Decision AI Analysis

**Application**: Small business loan, $150k, restaurant

**AI Analysis**:

## Creditworthiness Assessment
**Positive Factors**:
- Business credit score: 720 (Good)
- Personal credit score: 680 (Fair-Good) 
- 3 years operating history with consistent revenue
- Debt service coverage ratio: 1.4x (Adequate)
- Industry experience: 8 years in restaurant management

**Risk Factors**:
- High industry risk (restaurant failure rates)
- Seasonal revenue patterns identified
- Limited collateral beyond equipment
- Recent increase in business debt obligations

## Recommendation Framework
**Suggested Action**: Conditional approval with enhanced monitoring
**Recommended Terms**: 
- Loan amount: $120k (80% of request)
- Interest rate: Prime + 3.5% (risk-adjusted)
- Term: 5 years with personal guarantee
- Requirements: Monthly financial reporting, minimum DSCR covenant

## Compliance Documentation
- ✅ All factors based on creditworthiness criteria
- ✅ No prohibited basis factors considered
- ✅ Risk assessment based on verifiable data
- ✅ Terms consistent with bank's pricing policy
- ✅ Adverse action disclosures prepared if denied

**Required Review**: Senior credit officer approval for final decision

# Compliance Monitoring Framework

## Daily Monitoring
- AI system performance metrics
- Error rate and accuracy measurements
- Decision audit trail verification
- Regulatory limit compliance checks

## Weekly Reviews
- Model drift detection analysis
- Bias testing and fairness metrics
- Client complaint pattern analysis
- Supervisor review completion rates

## Monthly Assessments
- Comprehensive model validation
- Regulatory reporting preparation
- Risk metric trend analysis
- Staff training compliance verification

## Quarterly Audits
- External model validation review
- Regulatory examination preparation
- Policy and procedure updates
- Technology infrastructure assessment

## Annual Requirements
- Comprehensive risk assessment
- Regulatory filing updates
- Model governance review
- Third-party vendor assessments

# Decision Audit Trail Template
{
  "decision_id": "DEC-2026-03-31-001",
  "timestamp": "2026-03-31T10:30:00Z",
  "user_id": "advisor_123",
  "client_id": "client_456",
  "ai_model": "investment_advisor_v2.1",
  "input_data": {
    "risk_profile": "moderate",
    "time_horizon": "20_years",
    "current_allocation": {...}
  },
  "ai_recommendation": {
    "suggested_allocation": {...},
    "confidence_score": 0.85,
    "risk_assessment": "appropriate"
  },
  "human_review": {
    "reviewer": "senior_advisor_789",
    "approval_status": "approved_with_modifications",
    "modifications": "Reduced equity from 70% to 65%",
    "rationale": "Client age suggests slightly more conservative approach"
  },
  "regulatory_checks": {
    "suitability_verified": true,
    "fiduciary_standard_met": true,
    "disclosure_provided": true
  }
}

# Bias Testing Framework

## Prohibited Basis Testing
- Gender: Equal approval rates for similar creditworthiness
- Race/Ethnicity: Statistical analysis of outcome disparities  
- Age: Verification that age factors are legally permissible
- Geographic: Ensure redlining patterns don't emerge

## Statistical Testing Methods
- Adverse impact ratio analysis (80% rule)
- Regression analysis controlling for legitimate factors
- Matched pair testing with similar profiles
- Longitudinal monitoring of decision patterns

## Testing Frequency
- Pre-deployment: Comprehensive bias testing on training data
- Monthly: Statistical monitoring of decision outcomes
- Quarterly: Deep-dive analysis with external validation
- Annual: Comprehensive model audit including bias assessment

# Human Oversight Framework

## Tier 1: Automated Review
- Rule-based compliance checks
- Risk limit validation
- Data quality verification
- Exception flagging for human review

## Tier 2: Junior Staff Review  
- Standard decision verification
- Documentation completeness
- Client communication accuracy
- Escalation of complex cases

## Tier 3: Senior Staff Review
- Complex or high-value decisions
- Exception approval authority
- Model performance assessment
- Regulatory compliance verification

## Tier 4: Management Review
- Policy exception approvals
- Model validation oversight
- Regulatory examination support
- Strategic AI governance decisions

# Future-Ready Context Template

## Regulatory Environment
**Current Regulations**: [List current requirements]
**Pending Changes**: [Track proposed rule changes]
**International Considerations**: [Multi-jurisdiction requirements]

## Technology Evolution  
**Model Capabilities**: [Current AI capabilities and limitations]
**Integration Architecture**: [System integration patterns]
**Scalability Requirements**: [Growth planning considerations]

## Risk Evolution
**Emerging Risks**: [New risk categories and measurements]
**Mitigation Strategies**: [Adaptive risk management approaches]
**Monitoring Enhancement**: [Advanced detection capabilities]