The provided document, "STATE OF AI IN BUSINESS 2025" from MIT NANDA, presents preliminary research findings on the widespread yet largely unproductive adoption of Generative AI (GenAI) in businesses, terming this discrepancy the "GenAI Divide." The report highlights that despite significant investment, 95% of organisations see zero measurable return, with successful implementation often determined by the approach rather than technology quality. Key barriers include a learning gap, where systems fail to adapt or remember context, and an implementation chasm, where only 5% of custom enterprise AI tools reach production, while external partnerships are twice as likely to succeed as internal builds. Ultimately, the study advocates that crossing the divide requires buying customised, learning-capable systems that integrate deeply into workflows, focusing on back-office ROI and acting like strategic partners, rather than just purchasing generic software.