Phase 1: Preparation
1
Gather Credentials
Host, Port, DB Name, User with SELECT permissions.
2
Check Credits
Ensure you have sufficient balance (1 credit / 200 tables).
3
Plan Scope
Start with 10-20 tables to test before full run.
Phase 2: Connection Setup
Connect & Test
Navigate to Database Settings, select your DB type, enter parameters, and run a connectivity test.
Phase 3-4: Selection & Configuration
Dataset Selection
- All Tables: Comprehensive coverage
- Filtered: Specific schemas/patterns
- Manual: Select individually
Output Config
- Sample Rows: 5-10
- PII Detection: Enabled
- Confidence Scoring: Enabled
Phase 5: Generation
AI Processing
The ensemble model (Claude Opus + GPT-4o) analyzes metadata. Monitor progress in real-time (approx 2-5 mins per 10 tables).
Phase 6: Review & Refinement
Pass 1
PII Verification
Filter by 'PII Only'. Verify flags and remove false positives.
Pass 2
Low Confidence Review
Sort by confidence (low to high). Edit descriptions scoring <70%.
Pass 3
Spot Check
Review technical accuracy and terminology consistency.
Phase 7-8: Export & Distribute
Download Formats
Excel (Business), JSON (Devs), HTML (Wiki), PDF (Audit).
Distribution
Share Excel with analysts, commit JSON to Git, upload HTML to internal docs.
Success Criteria
- 100% table coverage
- PII accurately flagged
- Confidence scores reviewed
- Exported & Distributed