KYC Upgrade Guide
Complete guide for upgrading user KYC (Know Your Customer) levels to unlock higher transaction limits.Overview
OLIVE uses a tiered KYC system that balances ease of use with regulatory compliance. Higher KYC levels unlock greater transaction limits.KYC Levels
| Level | Requirements | Daily Limit | Monthly Limit |
|---|---|---|---|
| Basic | Phone + Name | 100,000 SLE | 500,000 SLE |
| Enhanced | ID Front + Back | 500,000 SLE | 2,500,000 SLE |
| Full | Verified ID + Address | 2,000,000 SLE | 10,000,000 SLE |
All new users start at Basic level after phone verification.
Upgrade Methods
- WhatsApp Agent
- Admin Dashboard
- API
Users can upgrade via the WhatsApp conversational agent
WhatsApp Upgrade Flow
The most common upgrade path uses the WhatsApp agent:Example Conversation
ID Requirements
Accepted Documents
| Document | Requirements |
|---|---|
| Sierra Leone National ID | Front and back photos |
| Passport | Photo page |
| Driver’s License | Front and back photos |
Photo Requirements
Do
- Clear, well-lit photo
- All text readable
- Full document visible
- Flat surface
Don't
- Blurry or dark images
- Glare covering text
- Cropped or partial
- Tampered documents
Validation Process
Multi-Layer Verification
The system uses multiple validation layers:1. OpenAI Vision
1. OpenAI Vision
- Validates document authenticity
- Checks for photo presence
- Verifies official document look
- Detects obvious tampering
2. OCR Extraction
2. OCR Extraction
- Extracts text using Tesseract
- Parses ID number, name, dates
- Calculates confidence scores
- Detects low-quality images
3. Cross-Validation
3. Cross-Validation
- Compares front and back data
- Verifies name consistency
- Checks dates match
- Flags mismatches
Extracted Fields
For Sierra Leone National ID:| Field | Source | Example |
|---|---|---|
| NIN (ID Number) | Front | SL-NIN-12345678 |
| Full Name | Front | John Doe Smith |
| Date of Birth | Front | 1990-05-15 |
| Gender | Front | Male |
| Date of Expiry | Back | 2028-03-20 |
Fraud Detection
The system automatically detects:| Indicator | Detection Method |
|---|---|
| Tampered text | OCR confidence analysis |
| Missing headers | Pattern matching for official text |
| Suspicious patterns | Keywords like “sample”, “test”, “photoshop” |
| Repeating ID numbers | Pattern detection (e.g., 11111111) |
| Front/back mismatch | Cross-validation of names/numbers |
| Low image quality | Confidence score thresholds |
Fraud Response Example
API Integration
For third-party integrations that need to submit KYC documents:Upload Single Image
Complete KYC Upgrade
Response
Troubleshooting
”Image validation failed”
| Cause | Solution |
|---|---|
| Blurry image | Ask user to retake in good lighting |
| Glare on card | Tilt card to avoid reflections |
| Partial image | Ensure entire document is visible |
| Wrong format | Convert to JPEG or PNG |
”Documents don’t match”
- Front and back should be same document
- Names should match exactly
- Dates should be consistent
”OCR confidence too low”
- Image may be too dark or have shadows
- Text may be obscured or damaged
- Ask for better quality photo