iKala Explores Agentic Commerce, GEO, and Open-Source Benchmark TMMLU+
Frontier AI models · AI assistants · Enterprise AI
iKala executives highlighted that shifting search behaviors are driving marketing from SEO toward generative engine optimization (GEO) and agentic commerce. To test local context comprehension, the team released open-source Traditional Chinese benchmark TMMLU+ v1.1 to evaluate AI models on Taiwan-specific cultural knowledge.
iKala co-founder and chairman Sega Cheng and leadership noted that as consumer decision-making and search gateways transition to AI, traditional SEO is evolving into generative engine optimization (GEO). Cheng stated that AI agents are advancing from information retrieval to executing decisions and transactions, driving agentic commerce and requiring brands to secure visibility on AI recommendation lists. Regarding internal enterprise adoption, Cheng pointed out that hurdles often arise when executives do not actively use AI, advocating an agile strategy to rapidly validate key scenarios. Addressing model localization, iKala developed and released the open-source Traditional Chinese benchmark TMMLU+ v1.1 in September to objectively assess model comprehension of Taiwanese cultural context and local knowledge.
Key takeaways
- iKala's Sega Cheng noted search gateways are shifting to AI, moving marketing focus from SEO to GEO to enter AI recommendation lists.
- AI agents are advancing from providing information to executing transactions, accelerating the emergence of Agentic Commerce.
- Enterprise AI adoption hurdles often stem from disengaged leadership, making agile validation across key use cases advisable.
- To assess model comprehension of Traditional Chinese and Taiwan-specific knowledge, iKala released open-source benchmark TMMLU+ v1.1 in September.
Source：數位時代 Read the original ↗
Prepared, checked and automatically published by AI; not individually reviewed by a human. Translation checked by AI.