Fragmented AI capabilities
Five different providers split core chatbot capabilities across incompatible platforms.
AI Chatbot SaaS Service for LG Group & B2B Partners
LG CNS Singlex Chatbot Platform
Consolidating fragmented chatbot operations into one enterprise SaaS builder
I led the UX project for an AI Chatbot Builder within LG Group's cloud SaaS platform, Singlex, in response to market demand for chatbot adoption and the group's goal of integrating AI services across its businesses.
- Team & duration
- LG CNS cross-functional SaaS team · 6 months
- Company
- LG CNS
Project scope & backgroundOpen details
I led the UX project for an AI Chatbot Builder within LG Group's cloud SaaS platform, Singlex, in response to market demand for chatbot adoption and the group's goal of integrating AI services across its businesses.
CompanyLG CNS ↗
Role / SkillsLead Product DesignerBot Migration Project Manager
DeliverablesService StrategySaaS Platform UX DesignService ArchitectureAI Chatbot Service Migration
Project dateAug 2020 - Jan 2021 · 6 months
Context
Context & Background
Five separate chatbot platforms fragmented AI capabilities and increased operating costs
Internal LG teams and B2B partners needed one cloud-based platform instead of repeatedly funding and operating disconnected chatbot systems.
Redundant investments
Each independently operated chatbot repeated product, integration, and maintenance work.
Infrastructure costs
The absence of a shared cloud environment increased operating and scaling costs.
Role & Decisions
Ownership
The decisions behind the product direction
- Mandate
- Lead the product experience while coordinating migration into one enterprise chatbot builder.
- My decision
- Unified build, test, access control, training, and migration while retaining essential Dialogflow compatibility.
- Trade-off
- Standardized core operations without breaking essential Dialogflow-dependent workflows.
- What I led
- Lead Product Designer · Bot Migration Project Manager. Service Strategy, SaaS Platform UX Design, Service Architecture, AI Chatbot Service Migration.
Hypothesis · HMW Approach
“How might we integrate each subsidiary's and B2B partner's legacy chatbot systems and requirements while making chatbot building and management more seamless in the cloud?”
We interviewed chatbot operators from each subsidiary and B2B partners including KB Kookmin Bank and GS Shopping, then benchmarked Google Dialogflow, Kakao i Connect, and Naver Clova.
Because some subsidiaries depended on core Dialogflow capabilities, we collaborated with Google to embed a Dialogflow mode in the platform.
Research & Strategy
Outcomes
Singlex AI Chatbot Builder
A low-code builder that enables teams to create chatbots without additional programming.
Product Evidence
Integrated Service Portal
A consistent portal across Singlex AI solutions helps users navigate directly to the service they need.
Chatbot Creation and Management Access Control
Permissions allow different teams and organizations to operate independently. A builder can be created and managed at company, organization, or team level.
Easy Build, Test and Management
Traditional builders required code for response conditions and offered limited options to operators. Singlex uses a GUI for conditions, APIs, and UI components, supports instant testing, and includes a debug mode that improves communication with developers.
Train the model and connect the service with low code
Operators connect services and additional data using developer-defined API rules. Real-time interaction monitoring exposes abnormal responses so teams can correct and retrain them quickly.
User Feedback & History Search
Negative feedback is most useful when its surrounding context is visible. The history page shows the user's input, the response, and the actions or buttons that preceded abandonment or feedback.
Impact
Impact
Consolidated platforms reduced cost and supported B2B SaaS revenue
The unified platform also supported delivery to B2B customers such as LX Pantos.
Chatbot systems unified
LG Innotek, LX Pantos, and other LG-group chatbot systems migrated to Singlex.
Response accuracy improvement
Google BERT NLP and intent/entity classification outperformed the previous Dap Talk builder in bulk testing.
More efficient operations
The core build-and-operation team was reduced from six specialist roles to four.
Monthly operating-cost reduction
Six decentralized chatbot systems were replaced with one SaaS platform.
Sales in the launch quarter
Chatbot transitions for LX Pantos, LG Innotek, and other partners supported SaaS revenue.