According to Ethan Mollick's latest AI tool guide, the core model of AI applications has shifted from purely conversational interaction to agentic systems over the past year. Mollick's guide initially focused on chat models like ChatGPT, Claude, and Gemini, supplemented by alternative modes like Deep Research, but now emphasizes agentic systems that 'can complete the equivalent of several hours of human work in one go.'
This shift reflects the industry's higher expectations for AI capabilities—no longer satisfied with simple Q&A, but pursuing autonomous execution of complex tasks. The update of Mollick's guide itself confirms this trend: he no longer lists Gemini as a recommendation because Google 'still doesn't have a clear product' in this category.