A new report from MIT’s NANDA initiative reveals that 95% of generative AI pilot projects in corporations fail to deliver meaningful revenue growth. Successes are primarily seen in startups focusing on targeted pain points and in companies adopting vendor partnerships. The prevalent issue is a learning gap in effectively integrating AI into existing workflows, rather than deficiencies in AI models themselves. The study highlights greater ROI in automating back-office operations compared to sales and marketing applications, underscoring the need for strategic AI adoption in enterprises.