Status: Delivered. Described from our own delivery records. Client identity, brand and commercial terms are withheld. Figures, where given, cover the period stated and nothing beyond it.
Context
A healthcare provider delivering care both on its own premises and in the community. Its workforce mixes employed support staff and volunteers, with different availability rules, qualifications and constraints.
Its funding comes through a government scheme that requires activity to be evidenced in a specific format.
The problem
- Shifts had to reconcile staff availability, volunteer availability, qualification requirements and patient need — a combination too complex to hold in a spreadsheet.
- On-site and off-site care were scheduled differently but had to appear as one picture.
- Activity was recorded separately from the schedule, so the two disagreed.
- Preparing the funding submission meant reconstructing what happened from several sources.
- A scheduling error was not just an inconvenience; it could mean a care visit not happening.
What we built
An AI-supported scheduling and shift-management system
Proposing assignments against availability, qualification and location constraints, with a human making the final decision.
Full activity tracking
Tied to the schedule, so what was planned and what happened are the same record rather than two.
Compliant billing exports
In the format the national funding system expects, generated from the activity record rather than assembled by hand.
How we delivered it
- 1
Map the constraint set
Who can do what, where, when, and under which rules.
- 2
Foundation
Data model for staff, volunteers, patients, sites and activities.
- 3
Scheduling core, then the assistive layer
With proposals always reviewable before they take effect.
- 4
Activity tracking
Wired so completion updates the schedule rather than a parallel log.
- 5
Billing export
Validated against real submissions before go-live.
- 6
Pilot on one team
Through a full funding period.
How it was phased
Indicative phasing for work of this shape. Actual duration varies with data quality, access and decision speed.
What the client keeps
- The production system
- The constraint and qualification model
- The billing export specification and its validation cases
- Source in their own repository
- Operational runbooks