KOTRA Navi: A New Benchmark for AI-Powered Qualified Buyer Discovery
KOTRA Navi is an AI-based qualified buyer discovery service developed under DX Innovation Lab Project 21. It transforms a traditionally labor-intensive buyer search process into an automated workflow, recommending the most suitable buyers for each company from a global database of more than 800,000 records while also verifying genuine purchase intent through AI analysis. The system combines local intelligence from 127 overseas trade offices with AI-driven assessment.
Development Process: From Problem Definition to Service Launch
The development of KOTRA Navi began with a practical field-level problem. Trade office staff typically spent two to three weeks identifying buyers manually, yet only about 15% of those leads resulted in actual transactions. To address this inefficiency, the project moved through a three-phase development process centered on AI automation.
AI Architecture: A Three-Layer Verification System
KOTRA Navi identifies qualified buyers through a three-layer verification system. Data-based automated checks, AI prediction models, and local trade office verification are applied sequentially to minimize false positives and improve lead quality.
| Verification Stage | Method | Assessment Criteria | Accuracy |
|---|---|---|---|
| Stage 1: Automated Data Check | Analysis of financial, transaction, and credit data | Revenue, import history, credit grade | 75% |
| Stage 2: AI Prediction | Ensemble ML model scoring | Composite score from 35 variables (0-100) | 88% |
| Stage 3: Local Verification | On-site visits and phone validation by trade office staff | Actual purchase intent and decision-making authority | 95%+ |
Application to Buyer Discovery in Bangladesh
KOTRA Navi is directly linked to the Dhaka Trade Center buyer database and provides features tailored to buyer discovery in Bangladesh. Its database includes more than 3,000 local buyers, and when combined with on-the-ground verification by trade office staff, it delivers a high level of screening accuracy.