University of Hertfordshire Higher Education Corporation and Delight Supported Living KTP 24_25 R2

Home care schedulers currently spend hours manually matching carers to clients, a process prone to errors that leave vulnerable people without visits. This project builds an AI system that automatically optimises routes and schedules in real time, accounting for changing client needs, carer skills, travel times, and shift preferences. The core problem is that existing scheduling software requires heavy human oversight, so mistakes cascade—a missed medication round or a late morning call can disrupt an entire day’s care for someone housebound. The AI will learn from live data, adjusting assignments as conditions shift, such as when a carer calls in sick or a client’s appointment runs long. If successful, the system could reduce administrative workload for care agencies, cut travel costs, and—most importantly—ensure that the most vulnerable receive consistent, timely care. The project is applied, not fundamental science: it adapts existing route-optimisation algorithms to the messy, human realities of domiciliary care, where a “job” is a person who depends on a knock at the door.