A joint report by Wharton and Accenture analyzed more than 18,000 tasks across 923 occupations and 19 industries, finding that more than 90% of tasks that AI agents can undertake can use cheaper, less capable models. Only 6%–11% of the associated work typically requires frontier models, while token costs account for just 20–33% of total AI agent system costs, with the remainder coming primarily from orchestration, human oversight, and integration. The report also found that fewer than one-third of organizations have achieved sustained executive-level business value from AI agent initiatives.
Source:Crypto Briefing · Read Original
Leave a Reply