# AI Features Scoping Plan

## Context
Board wants to scope potential paid AI features for Lendair. Goal is monetization via premium AI features on top of core micro-lending platform.

## Proposed AI Features (v1 Paid Tier)

### 1. Smart Loan Matching (Lender)
**What:** Match available loans to lender preferences on returns and risk.
**User value:** Saves time, surfaces best-fit opportunities automatically.
**Data needed:** Loan terms, lender risk appetite, historical borrower performance, return projections.
**Complexity:** Medium. Needs preference engine + scoring model.

### 2. Credit/Trust Score (Both Sides)
**What:** AI-generated score for each party's loanworthiness/history. Lenders see borrower score; borrowers see their own lender reliability score.
**User value:** Trust signal reduces friction on both sides.
**Data needed:** Historical repayment data, loan completion rates, on-time payment history, defaults.
**Complexity:** Medium. Requires historical data accumulation; harder for new users.

### 3. Loan Health Dashboard (Borrower)
**What:** AI-powered view showing repayment trajectory, early warning for cash flow issues, suggested prepayment amounts.
**User value:** Helps borrowers stay on track and build credit.
**Complexity:** Low-Medium.

### 4. Risk-Adjusted Return Engine (Lender)
**What:** For each loan, show projected return after historical default probability. Helps lenders optimize portfolio allocation.
**User value:** Better risk/return tradeoff decisions.
**Complexity:** Medium.

### 5. Auto-Diversification Advisor (Lender)
**What:** Suggest optimal loan distribution across borrowers to minimize portfolio risk.
**User value:** Professional-grade portfolio management.
**Complexity:** Medium-High.

### 6. Fraud/Anomaly Detection (Platform)
**What:** Detect suspicious borrowing patterns, synthetic identities, unusual default clustering.
**User value:** Platform safety, protects lenders.
**Complexity:** High (but defensive - could be free feature).

## Prioritization for MVP Paid Tier
1. **Loan Matching (Lender)** - Board specified as obvious first win
2. **Trust Score (Both)** - High trust signal, good differentiator
3. **Risk-Adjusted Returns** - Natural extension of #1

## Revenue Model
- Subscription: bundle v1 features, charge ~$5-15/month
- Or: per-feature unlock

## Next Steps
1. Board confirms direction
2. Hand off to CTO for technical feasibility/estimation
3. Create product spec issues for prioritized features