Overview
"Next-generation Financial Services powered by AI Agent: Digital Consultants that Communicate Like Humans"
The LG AI Digital Human Kiosk is an AI Agent system that integrates multiple AI technologies to naturally communicate with customers like real bank consultants.
We have implemented an intelligent kiosk that understands and responds to customer inquiries in real-time by integrating STT (Speech Recognition), LG EXAONE LLM (Conversational AI), TTS (Text-to-Speech), A2F (Facial Animation), and INFO UI. Combined with hardware such as cameras, directional microphones, and proximity sensors, it provides an environment where customers can naturally approach and converse through voice.
Designed to enable AI to perform complex financial tasks such as loan and financial consulting, it serves diverse customer segments through multilingual support. Beyond simple information provision, it operates as a true AI Agent that understands customer situations and proposes customized financial solutions.
Client
LG
Duration
3 Months
Technology
Multi-AI Integration
Type
AI Agent Kiosk
Core Technology
- • STT (Speech-to-Text) - Accurate collection of customer inquiries through real-time speech recognition
- • LG EXAONE LLM - Professional consulting using finance domain-specialized large language model
- • TTS (Text-to-Speech) - Friendly response delivery through natural voice synthesis
- • A2F (Audio-to-Face) - Real-time facial animation synchronized with voice
- • INFO UI - Enhanced visual understanding through intuitive information display
- • Multilingual Support - Language processing system capable of serving global customers
Key Features
- • Natural conversation experience through real-time voice recognition and response system
- • Smart detection using cameras, directional microphones, and proximity sensors
- • AI Agent functionality specialized in loan and financial consulting
- • Immersive digital human visuals that express emotions
- • System architecture perfectly integrated with banking business processes
- • Efficient integration of multiple AI models with optimized response speed
