LAP (Lifegraphy)
Synthetic consumer agents from a 4,000-respondent survey — four-month deployment with 47 practitioners at LG Electronics
The High-Stakes Context
For years, businesses have relied on traditional methods like personas to grasp their target audience — valuable, but often time-consuming, subjective, and static. Lifegraphy is a Synthetic Customer Agent (SCA) system that transforms how enterprises conduct customer research: an interactive chat system generating dynamic agents from a massive lifestyle survey of 4,000 respondents. It combines visual cluster profiles, advanced SCA generation, and interactive chat rooms that let practitioners converse with representations of their ideal customers in real time.
The Architectural Orchestration
Architected an LLM pipeline incorporating Retrieval-Augmented Generation (RAG) and conversational memory to synthesize digital agents that flawlessly execute role-playing scenarios (e.g., "Smart Innovator", "Tasteful Connoisseur"). Orchestrated the system to integrate keyword frequency mapping and emotion analysis.
The Deployment & Scale
Deployed the fully-functioning product-level system "in the wild" to 47 global enterprise practitioners at LG Electronics for 4 months.
The Empirical Evaluation
Synthesized system logs, surveys, and interviews across the deployment. 78.6% of practitioners agreed the agents can complement interviews and surveys, 69.1% found responses detailed enough to support expert tasks, and 64.3% would recommend the system to a colleague.
The Outcome
- Four-month internal deployment at LG Electronics — presented to LG Chairman
- Used across six departments for customer research
- Research paper in preparation (co-first author)