A recent survey by Bain & Company reveals a significant disconnect between AI investment and realised cost savings. The study, which polled 951 organisations, found that nearly 40 percent of firms achieved less than 10 percent in savings, falling short of the typical target of 11 to 20 percent. Despite this underperformance, nine out of ten companies intend to increase their spending, particularly on AI agents. The primary obstacle identified is excessive human involvement in workflows that were designed for full automation. Only 7 percent of respondents operate fully autonomous agents, while 38 percent require human approval for every step, and another 32 percent intervene only when necessary. This friction prevents the efficiency gains promised by initial business cases.
The findings highlight a critical operational gap where process design fails to match technological capability. Data access remains the second largest hurdle, cited by 41 percent of participants, suggesting that technical infrastructure is not the sole barrier but rather a symptom of broader management issues. Bain advises leaders to treat data governance as a strategic priority rather than an IT task, urging firms to restructure workflows before deploying AI tools. The persistence of human-in-the-loop models indicates a cultural resistance to true automation, limiting the scalability of AI initiatives. Until organisations align their internal processes with the autonomy levels required by modern AI, financial returns will likely remain elusive despite growing budgets.
- Nearly 40 percent of surveyed companies failed to meet their AI cost-saving targets of 11 to 20 percent.
- Excessive human oversight, with 38 percent requiring approval for every action, is the main cause of lagging efficiency gains.
- Leaders are urged to treat data access and process redesign as management responsibilities rather than purely technical challenges.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
