Sompo Holdings and NTT signed an agreement on 11 November 2026 to study the use of AI-equipped robots and connected systems in Japan's nursing-care sector. Announced the following day, the partnership starts with exploring applications, not a completed deployment.
The expected initial tasks include cleaning, laundry, preparing and clearing meals, and transporting items, rather than providing direct physical care to residents. The distinction puts the work surrounding care at the centre of the study, while leaving specific applications and demonstration plans to be defined.
Chores Are the Initial Target
The partners describe staff shortages as an intensifying challenge for Japan's nursing-care services. Their stated aim is to ease caregivers' workloads while maintaining or improving user-centred care, making the practical value of the proposed systems a matter of staff time as well as technical capability.
The announcement uses 'physical AI' to cover cameras, sensors, robots and the systems connecting them. These technologies perceive their surroundings and use that information to perform physical tasks. For carers, the intended benefit is help with routine work; for residents, preserving care quality remains part of the objective, not an outcome the agreement has already delivered.
Earlier Robots Added Work for Carers
Earlier experience shows why assigning work to a robot does not automatically lighten a carer's day. In October 2025, the BBC reported observations by researcher James Wright, who spent seven months studying robots in a Japanese care home. Cleaning, recharging and troubleshooting the machines consumed staff time.
'After several weeks the care workers decided the robots were more trouble than they were worth and used them less and less, because they were too busy to use them', Wright told the broadcaster. Those robots supported lifting, therapeutic interaction and exercise, rather than the initial chores proposed under the new agreement.
The comparison concerns the demands of operating machines in care settings; it does not establish how the proposed systems will perform or whether they will produce the same difficulties for staff.
The Study Must Establish Workload Benefits
The current partnership explicitly includes operational design, maintenance and security alongside technology development. Sompo will contribute knowledge of frontline care and service improvement, while NTT brings communications, device-integration and operational expertise.
Their proposed roles connect the requirements of care services with the infrastructure needed to keep equipment working.
The companies will first identify frontline challenges and priority uses, then consider how people and physical AI would work together. Target applications, demonstration plans, data and intellectual-property arrangements, and commercial responsibilities still need to be defined.
The agreement therefore establishes a programme of work rather than evidence that these systems already save carers time.
The partners also plan to explore collaboration with physical AI business operators. Their study covers the route from research and demonstrations through deployment, continuing operation and improvement, rather than stopping at a technical demonstration.
A machine completing a delivery or cleaning task would establish a capability, but not by itself demonstrate a lighter staff workload. The operational demands highlighted by earlier experience belong in that assessment too: the intended benefit depends on reducing the work surrounding care, not simply moving tasks from people to machines.
