How a UK home healthcare provider cut cost per case by 35% with Noxus

A multi-region care provider in the UK managing urgent inbound referrals from families and hospitals across multiple locations

At a glance…

83%of communications handled automatically

35%reduction in cost per case

70–85%of referrals processed without manual triage

23 daysto first AI Worker live

A multi-region care provider with intake as the bottleneck

A multi-region domiciliary care provider in the UK managing urgent, complex inbound enquiries. Families seeking immediate care. Hospitals coordinating patient discharge. Every enquiry required structured operational checks before any acceptance could be given. All handled manually.

Unstructured referrals, manual triage, and winter surges

Inbound referrals arrived by email, unstructured and incomplete. Coordinators reconstructed care needs from fragmented messages, chased missing information across multiple follow-ups, and manually cross-checked capacity with scheduling teams. Urgent hospital discharge cases competed with routine private enquiries in the same shared inbox.

Every referral required the same checks before acceptance: postcode coverage, rota capacity, care complexity, funding confirmation, and whether managerial approval was needed. None of it was automated.

During winter pressures and discharge surges, temporary admin staff were brought in just to manage intake volume. The bottleneck was not care delivery. It was referral intake.

One AI Worker across email and web referrals, live in 23 days

The provider deployed a Noxus AI Worker to manage referral intake and qualification across email and website forms.

Unstructured inbound emails are extracted, standardised, and triaged automatically. When information is incomplete, a single structured checklist is sent — replacing repeated back-and-forth. Capacity and coverage are validated automatically against live data. Complex or urgent cases are flagged and routed to managers with full context, no manual triage required.

Final acceptance decisions remain human-led throughout.

The first AI Worker went live in 23 days.