A new report from the Massachusetts Institute of Technology reveals that 95% of companies investing in artificial intelligence are failing to generate tangible results, a phenomenon researchers term the “GenAI Divide”. The finding lands just as the technology itself undergoes a structural shift: the emergence of agentic AI, systems that no longer merely respond to prompts but plan, decide and execute multi-step workflows autonomously. Latin American business leaders and global technologists describe this as a second wave, distinct in nature from the chatbot era, and one that demands organisations rebuild their foundations before chasing advanced models.

Across Mexico, executives from Dell Technologies, Microsoft and steelmaker TYASA stress that AI projects fail not because of the technology but because of people and processes. TYASA spent months evaluating alternatives before adopting a hybrid cloud infrastructure, a move that cut licensing costs by 30% and improved response times. The company’s IT manager, José Guerrero Corona, notes that AI is only as good as the data it receives, and that clarifying business objectives and organising information must come first. This view is echoed in Brazil, where industrial maintenance platform Melvin has embedded AI directly into asset management. Its tools calculate equipment reliability, auto-schedule maintenance teams and detect early failure signals from vibration and temperature sensors. Co-founder Eymard Barroso says the technology only makes sense when it generates concrete operational results, a principle validated during the firm’s recent recertification under ISO 9001, 27001 and 27701.

In Argentina, a joint initiative by software unicorn Globant and learning platform Egg targets the human dimension of the gap. Their “AI Talent Shift” methodology, already applied to 60,000 employees at Mercado Libre, measures emotional readiness, capability alignment and peer influence through a proprietary Initial Sync Index, rather than static surveys. Ignacio Gómez Portillo, Egg’s CEO, argues that 70% of the value-capture challenge resides in people and processes, not technology. The programme reports an 80% behaviour change rate, underscoring a broader Latin American consensus: successful adoption requires synchronising the entire organisation, not just deploying licences.

Viewed from Southeast Asia, the conversation extends to higher education. Xi’an Jiaotong-Liverpool University, a China-UK joint venture, has made AI a compulsory subject from the first year, embedding it across disciplines under a “X plus AI” philosophy. Executive President Youmin Xi insists that mastering AI tools is insufficient; graduates must understand context, risk and ethical limits. The curriculum progresses from foundational use to discipline-specific application and finally to AI-assisted research, positioning the technology as a partner rather than a replacement.

The next milestone to watch is the scaling of agentic architectures beyond pilots. As autonomous agents begin coordinating financial risk analysis, contract drafting and industrial maintenance, the pressure on governance frameworks and workforce reskilling will intensify. The MIT data suggests that without deliberate organisational transformation, the gap between AI’s promise and its practical returns will widen, even as the technology itself grows more capable.