AI is reshaping work, but businesses are slow to adapt

AI is making businesses more efficient, but so far it has had a limited impact on how organisations operate. This is according to Deloitte’s international study The State of AI in the Enterprise 2026.

For the study, Deloitte surveyed 3,500 business leaders across 25 countries. Although 81 percent say AI has improved the speed of decision-making and execution, only 25 percent use the technology to fundamentally transform their business operations. Organisational change is therefore lagging behind technological progress.

According to Deloitte, the challenge is no longer the availability of AI, but adapting workflows, business models and decision-making processes to make better use of its potential.

AI Is changing the way people work

Sixty percent of the organisations surveyed have redesigned jobs to some extent or to a moderate degree around AI. However, extensive changes remain rare: only 15 percent report making substantial or complete adjustments. Just 15 percent also say AI has led to significant or fundamental changes in decision-making.

One example in the report is a well-known furniture retailer that introduced an AI tool allowing customers to design rooms. The technology did not replace human designers, but changed their tasks. After receiving additional training, they took on a different role and were able to add more value. One executive described this as optimising the workforce, rather than reducing it.

At other organisations, AI helps employees work faster, but they still follow the same approval procedures. In these cases, the technology speeds up existing processes without fundamentally changing them. Deloitte therefore argues that businesses should look beyond which tasks they can automate and consider how work, responsibilities and decision-making authority could be reorganised.

Maintaining control over AI

As AI systems become more autonomous, clear rules on accountability are becoming increasingly important. Forty percent of organisations allow AI to make decisions or take action independently, to some or a moderate extent, without human approval. Yet at 23 percent of organisations, it is unclear who is responsible when an AI system produces a negative outcome.

This calls for oversight that reflects what a system is allowed to do. An AI tool that only makes recommendations requires different controls from a system that independently takes action. Businesses should therefore establish who is responsible, when human approval is required and when a system must refer a decision to an employee.

The underlying technology and infrastructure also play a role. For example, 74 percent of organisations consider it important for their data and computing power to remain available within specific geographical regions. In addition, 62 percent are concerned about their dependence on foreign AI technology and computing providers. Deloitte refers to this as Sovereign AI: maintaining digital autonomy and limiting geopolitical dependencies.

To retain control, organisations are exploring their own AI capabilities, regional suppliers and private cloud or on-premises infrastructure. Decisions about technology, infrastructure and oversight are consequently becoming increasingly important to business strategy.

Physical AI is growing faster than expertise

Software-based AI is not the only technology gaining ground in businesses. Physical AI, which combines AI with sensors, robots and intelligent machines, is also becoming more widespread. Currently, 25 percent of the organisations surveyed use this technology to some or a moderate extent. That figure is expected to rise to 48 percent within two years.

One example in the report is a European chemical company that uses a robot to monitor factory operations. The robot can determine its position independently and interpret its surroundings, but does not fully replace employees. An infrastructure company also opts for human-machine collaboration: it automates repetitive tasks while an employee remains responsible for approval.

The growth of Physical AI does, however, expose a shortage of knowledge and skills. Only 19 percent of companies rate their own expertise as high or very high. The technology requires knowledge of robotics, hardware integration and safety. Companies that already have this expertise are leading the way: 64 percent use Physical AI to some or a moderate extent, while 28 percent have extensively deployed or fully integrated the technology.

Getting more value out of AI

The findings point to the same challenge: businesses are using AI more widely, but the organisational changes needed to make the most of the technology are lagging behind.

According to Deloitte, companies need to get more value from AI by redesigning work and decision-making responsibilities. This requires insight into AI costs, links to commercial results and appropriate governance. Businesses must also build expertise in Physical AI and maintain control over their data, infrastructure and suppliers. This will enable them to scale AI safely and make better use of it to drive growth and innovation.