AI Chatbot “Askonomy” for Vietnamese Economic Journalism
John Harding
English Editor-in-Chief, Vietnam Economic Times
"I highly appreciate the excellent translation quality of the Askonomy platform. Thanks to it, my work efficiency has improved significantly......"
The Client: Vietnam Economic Times/VnEconomy
Vietnam Economic Times/VnEconomy, a leading media and economic research organization, partnered with XNOR Group to launch Askonomy—an AI assistant that redefined how journalists and readers access economic data. This project set new standards for speed, accuracy, and newsroom efficiency in Vietnam.
AI assistant for Vietnamese economic journalism
Discover how Vietnam Economic Times/VnEconomy partnered with XNOR Group to pioneer an AI chatbot for Vietnamese economic journalism, optimize newsroom operations, and enhance user experiences.
Project Summary
Amid the global rise of Generative AI, VnEconomy took the lead in March 2023 by launching Askonomy, an AI chatbot for Vietnamese economic journalism, well ahead of global outlets such as the Financial Times and BBC. XNOR Group guided the entire journey, from market research and system architecture to optimizing AI models for Vietnam’s economic context. Dr. Trung Huynh, founder of Actable AI and now Head of AI at XNOR Group, led the technical development.
In March 2024, VnEconomy advanced further by releasing a proprietary language model version of Askonomy, developed with XNOR Group’s expertise. This upgrade reduced operational costs, improved translation accuracy, and gave the newsroom full control over its technology. Today, Askonomy delivers thousands of monthly visits, sub-second response times, and higher-quality content production.
Client Challenges
Technology and operational barriers hindering newsroom development:
-
Fragmented Data: Economic information scattered across portals slows research.
-
Language Barriers: Manual translation of economic terms is slow and error-prone.
-
Time-Intensive Research: Journalists spend excessive time gathering basic data.
-
Limited Analysis: Historical and current data can’t be easily correlated.
-
Quality Control Issues: Inconsistent translation and interpretation cause delays.
-
Access Needs: Voice query required for mobile or phone-based reporting.
Our Comprehensive Solution
XNOR Group designed and deployed an AI chatbot combining expert consulting, proprietary AI models, and streamlined newsroom workflows.
Breakthrough Features
-
Proprietary Language Model: Tailored to Vietnamese economic terminology.
-
Authoritative Data Sources: Built on VnEconomy’s 20+ year content repository.
-
Deep Search: Multi-query search ensures the most precise results.
-
AI Text-to-Speech: Summarizes and translates text for flexible use across environments.
Specialized Capabilities
-
Intelligent Q&A: Understands and answers complex, multi-topic economic questions.
-
Multilingual Translation: High-accuracy Vietnamese ↔ English translation for financial reporting.
-
Content Summarization: Condenses long reports into digestible insights.
-
Data Visualization: Presents economic data in charts for easier comparison.
-
Continuous Updates: Automatically refreshes with VnEconomy’s growing database.
Development and Deployment Process
Market Needs Assessment
- Detailed survey of journalist and editor requirements
- Analysis of technology gaps in Vietnamese economic media
- Competitive assessment against international solutions
Technical Research
- Analysis of AI/NLP models suitable for Vietnamese language
- Evaluation of data sources and information quality from VnEconomy
- Optimal system architecture design
UX/UI Design
- Wireframe and prototype creation based on editorial team feedback
- User-friendly interface design for journalism professionals
- Experience optimization for both desktop and mobile platforms
Backend Development
- API layer construction with Node.js and Express.js
- NLP model integration and natural language processing
- MongoDB database setup for storage and retrieval
AI/ML Integration
- Fine-tuning language models for Vietnamese economics
- Development of specialized translation modules
- Search algorithm optimization and result ranking
Quality Testing
- Comprehensive unit testing and integration testing
- Performance testing with large data volumes
- User acceptance testing with actual editorial staff
Performance Optimization
- Response time improvement to under 1 second
- Translation accuracy optimization to 95-99%
- Model fine-tuning based on real-world feedback
Production Deployment
- AWS deployment with microservices architecture
- Monitoring and alerting system setup
- Backup and disaster recovery planning
Training and Support
- Training sessions for editorial teams
- Detailed documentation and user manual creation
- Support channels and feedback system establishment
Proprietary Model Development
- Research and development of proprietary language models
- Migration from ChatGPT to self-developed models
- Operational cost optimization and enhanced security
Continuous Improvement
- Usage data collection and analysis
- Feature updates and improvements based on user feedback
- Processing and data analysis capability expansion
Project results
- Thousands of monthly visits
- <1s Response time
- Significant Work efficiency improvements
Customer Review
I highly appreciate the excellent translation quality of the Askonomy platform. Thanks to it, my work efficiency has improved significantly... Certainly, this is a specialized language model, which means the application's vocabulary (both in understanding and providing answers) is based on economic terminology. I am truly impressed with the translation quality and very pleased that the tool makes my work much easier.
John Harding
English Editor-in-Chief, Vietnam Economic Times
