AI Chatbot Projects for B2C Enterprise Partners
B2B2C Chatbot Projects
Turning partner VOC into a repeatable enterprise chatbot operating model
Providing 24/7 customer service became essential for B2C companies. Based on our AI-assistant operating experience, we delivered finance and retail projects using Kakao i Connect, NLP-based chatbots, and customer-service solutions.
- Team & duration
- Partner and Kakao cross-functional teams · 6 months per project
- Company
- Kakao Enterprise
Project scope & backgroundOpen details
Providing 24/7 customer service became essential for B2C companies. Based on our AI-assistant operating experience, we delivered finance and retail projects using Kakao i Connect, NLP-based chatbots, and customer-service solutions.
CompanyKakao Enterprise ↗
Role / SkillsProduct ManagerProduct / Conversation Designer
DeliverablesProduct PolicyUX DesignAPI ArchitectureConversation Design
Project dateFeb 2021 - Aug 2022 · 6 months for each project
Context
Context & Background
Customer wait times surged while demand for 24/7 connectivity increased
Pandemic closures disrupted contact centers and limited customer engagement. Demand shifted to KakaoTalk, South Korea's leading messenger with 50 million users, and partner-company apps.
Many partners lacked the expertise required to implement and sustain chatbot services.
Service Pressure
Customer-service wait time increased fivefold in one year
Pandemic-era contact-center disruption made always-on digital service a measurable operational need.
Increase in customer-service wait time
Source: Korea Federation of Banks
Role & Decisions
Ownership
The decisions behind the product direction
- Mandate
- Turn partner-specific VOC into a repeatable product and conversation-design operating model.
- My decision
- Used real VOC data and guided conversational UX to cover core intents while keeping operations sustainable.
- Trade-off
- Combined predefined high-confidence intents with generative coverage instead of relying on either approach alone.
- What I led
- Product Manager · Product / Conversation Designer. Product Policy, UX Design, API Architecture, Conversation Design.
Hypothesis · HMW Approach
“How might we deliver AI chatbots optimized for clients and ensure sustainable operations?”
A chatbot is not a one-time deliverable. Partner workshops defined each MVP, roadmap, and key feature set.
Because chatbots require continuous monitoring and learning, we designed a sustainable operating system and tools that clients could use to manage and maintain their service.
Research & Strategy
User Research
Stakeholder interviews across finance, retail, IT, and customer care
Interview participants
- NH Investment customer-care staff and IT engineers
- Korea Investment customer-care staff and IT engineers
- Kakao i Connect Skill Builder product owner
- GS Retail IT and infrastructure staff
- GS Retail customer-care staff
Voice of Customer data analysis
Average VOC records analyzed and preprocessed
User query types defined as intents to improve recognition
Utterances generated and trained for the on-premise NLU
Partial generative AI expanded coverage beyond the 30 predefined intents
Open-ended inputs were handled through small talk, web-based response generation, and a financial knowledge bot. Training data was generated from real VOC inputs.
Product Evidence
Outcomes
Kakao i Connect Chatbot
We built partner-specific chatbots by combining company requirements with a clear understanding of end-user needs.
Welcome Response User Experience
First-time users often do not know what the chatbot can do. A welcome message and card buttons expose available services and provide fast access to common tasks.
Guide User Experience
Ambiguous requests first receive a broad category response, followed by detailed guidance and examples that help users express a more accurate request.
Knowledge-search bot features using generative AI
Pandemic-era stock trading brought many new users with limited financial literacy and increased support demand.
A GPT-2-powered knowledge-search bot explained financial terms, IPOs, and other essential concepts, reducing support burden while improving financial literacy.
Impact
Impact
Six launches created measurable service and business value
The projects established a repeatable implementation and operating model for enterprise chatbots across finance and retail.