Multi-AI 아키텍처 (LG webOS26)

• Background Problem: Smart TVs have inconvenient input methods and vast content, causing a bottleneck in "search/exploration." There is significant room for AI intervention, but reliance on a single assistant/single model limits UX due to functional/language/policy/quality deviations. • Limitations of existing methods: (1) UX is dependent on the performance of a specific AI vendor (2) Personalization is enhanced, but privacy/security concerns are significant (3) "Contextual recommendation/setting automation," which is most important on TV, is fragmented.

• Provides access to multiple LLMs (Copilot, Gemini, etc.) at the OS level (integrated into the webOS layer, not an app-level function). • Designs personalization and security layers (mentioning LG Shield) based on TV usage data together, aiming to solve the personalization-security trade-off through product design.

• Adoption entities: LG TV users (final) and content/advertising/platform partners (indirect) who benefit from advanced recommendation/exploration. • B2C/B2B/B2G: B2C + B2B2C

• Environmental/regulatory constraints: Functional differences may exist by region depending on national personal information regulations and contracts/availability with LLM providers. • Industry/market scalability: Very high scalability based on webOS-搭載 TVs (strong "deployment" from an installed base perspective).

• CES Award Context Evaluation: Aligned with the core topics of home appliances in 2026 (multi-AI, personalization, security). • Technical Completeness: The structure is convincing, but the experience depends on operational quality, such as response speed, accuracy, and data notification/control. • Market Expectations: High expectations in the trend where OS/services become TV differentiators beyond panel competition. • Overrated/Underrated Factors: If "multi-AI" simply provides choices, the experience may be low, so the key is how deeply it is integrated into core user journeys.