Automating tasks versus redesigning work around AI

In 2026 BCG surveyed almost 12,000 employees, managers and leaders in more than twelve countries for the fourth time. Three quarters of front-line staff now use AI daily or a few times a week, 23 points more than in 2025. Of those who use it regularly, 42 per cent say they save eight hours a week, a full working day. Two thirds get little or no guidance on what to do with those hours, and more than half do not put them into other work. Meanwhile, the share of organisations using AI to redesign workflows end to end or to build a new business model rose from 22 to 42 per cent. These organisations do better on time saved and on the value it produces than organisations that merely roll out tools.

Mercer hears from more than 12,000 executives, HR leaders, investors and employees (surveyed in September and October 2025) that 98 per cent of executives want to change their organisation design in the next two years. For 63 per cent of the C-suite, redesigning work with AI and automation is the people investment with the best return, and 63 per cent say they need to move towards skills-based organisation. At the same time, 65 per cent expect between 11 and 30 per cent of the workforce to be redeployed or retrained because of AI, and employees’ concern about job loss rose from 28 per cent in 2024 to 40 per cent in 2026.

Between January and April 2026, JLL surveyed more than 2,200 executives and real estate leaders in 21 countries. Sixty per cent expect AI to redefine roles and 40 per cent expect it to replace them, and the same ratio holds for growth versus shrinkage of the workforce. Fifteen per cent of organisations have reached the stage of optimising AI; most are still at the stage of monitoring and analysing. The World Economic Forum sums up the experience of more than twenty large technology companies as follows: the gains from AI fail to materialise as long as leaders do not redraw the workflows, and their examples of large savings invariably come from a process that has been redesigned as a whole.

Across the four sources the picture is consistent. The intention to redesign work is almost universal, implementation has begun in fewer than half of organisations, and the hours gained evaporate as long as automation happens task by task rather than workflow by workflow.

In October 2026 Josh Bersin sees task analysis making a comeback. Vendors such as Lightcast, Draup, SAP and Workday are building large databases of tasks, so that jobs can be read as bundles of work an agent could take over. According to Bersin, task automation suits structured and regulated work above all, each task has to be weighed separately and a task list quickly goes out of date. Bersin therefore has organisation design start from the purpose and the outcome for the customer.

In automotive engineering, 83 per cent of the 800 executives surveyed by Capgemini Research Institute in July 2026 expect agentic AI to give engineers more time for core work and less time lost to admin and coordination. Laurence Noël, global head of automotive at Capgemini, says the organisations getting the most value redesign their workflows, strengthen collaboration across the value chain and set down who is accountable for decisions reached with AI.

How solid is the evidence? Four independent organisations measure the same shift within a single year, and BCG has a four-year series showing a doubling in twelve months. Most of the figures concern intentions and implementation is lagging, so the picture is not yet settled.