Overview

• Executives describe a shift to autonomous “agent” systems that can execute, measure errors and self‑correct, raising control and governance concerns as agents interact across APIs.
• SME experimentation is widespread, with IDC reporting about 70% in Argentina testing AI and a Microsoft/Edelman study showing rising priority and planned investment, as some analysts question ROI and Nvidia researchers argue many agents could run on smaller, cheaper models.
• A Salesforce Agentforce deployment at Andina ART cut a key process from 29 hours to 8 minutes on average with an 81% conversion rate, followed by a second tool that halved costs and hit 99% data quality after lessons from an initial failed implementation.
• Public trust remains fragile, with a KPMG–University of Melbourne survey finding 54% reluctant to trust AI, strong support for regulation and verification against misinformation, and high reported use by employees and students despite limited formal training.
• As 1.6 million students return in Catalonia, teachers and pupils are already using AI at scale, only 14% of European teachers expect learning gains, demand for training surged to 2,700 course requests, and schools weigh shifting assessment toward process and critical thinking.