CHALLENGES AND PROBLEM
- Diverse communication channels: Web, Facebook, Zalo, fanpage... information must be consistent
- Instant feedback & high accuracy: AI must respond correctly, clearly, and promptly.
- Handling content beyond the scope of AI: When encountering an unfamiliar question, the chatbot must transfer to a human agent.
- Keep system knowledge up to date: AI data needs to be continuously updated to keep up with reality.
SOLUTION
1. Implement external channels (customer-oriented)
- Integrate AI Chat into partners' official social channels (Website, Fanpage, Zalo OA, etc.).
- Answer customer questions about products, how to play, results, promotions, and general information about partners.
- Ensure fast, accurate, and consistent responses across all channels.
2. Implement an internal channel (Develop a virtual assistant to support customer service representatives)
- Develop a custom internal AI chat interface for the partner's customer service team.
- Allow staff to quickly look up information about products, how to play, legal regulations, procedures for handling situations, etc.
- Supports simulated conversations between customers and staff, helping to improve customer service skills.
- Serves as a centralized knowledge repository, providing accurate, consistent information for the entire internal system.
3. AI chat operation process

Step 1: Build the KMS data repository
Bellsystem24 Vietnam personnel upload data to the system, including: consulting scripts, product information, legal documents, etc.
Simultaneously, regularly update and continuously check the accuracy of AI in the customer response process, ensuring that responses are always based on accurate content from the KMS data repository.
Step 2: Customers interact through multiple channels and AI chat responds
Customers can interact with the AI chatbot through multiple channels (Website, Zalo, Facebook, etc.). The AI chatbot will respond based on the KMS knowledge base approved by the customer, using natural language established by Bellsystem24 Vietnam.
Customer service content support system
Note: Any questions beyond the scope of official data will not be automatically answered by the chatbot but will be transferred to Step 3 for processing.
Step 3: Forward to agents
When the AI chatbot lacks data or the customer requests it, the message will be automatically transferred to agents for direct support.
From there, support staff will continue to provide answers and may update the KMS data if necessary.
Step 4: Reporting and analysis
System records all interactions
Statistics: number of chat sessions, successful auto-response rate, forwarded sessions, topics of interest...
PROJECT RESULTS
- Processing capacity: 9,000 messages/month
- Customer experience is significantly improved
- Significant increase in work efficiency – reduced workload for call centers / customer service representatives
- Automation combined with human support when needed increases flexibility and efficiency.