PUB-010/Whitepaper/Aug 2026
A digital twin for a large organisation: deterministic simulation, grounded AI and spatial operations
Engineering case study on the NHS Digital Twin proof of concept, including procedural floor plans, navigable operational state and a grounded conversational interface.
Business value
The approach demonstrates how a large organisation can be understood as one navigable system before every live feed is connected. It can support discovery, operational training and integration planning while clearly separating simulation from telemetry.
Evidence and measures
- - The prototype validates zero unreachable room pairs and zero wall-crossing overshoot in its dedicated harness.
- - Carried assets tracked their porter with approximately 0.000 positional drift in the recorded validation.
- - Movement and occupancy are simulated; decision-time improvement with NHS operators remains unmeasured.
Research Questions
- 01Can procedural spatial models accelerate digital-twin discovery where complete BIM or telemetry is unavailable?
- 02How should grounded AI expose sources and uncertainty inside an operational map?
- 03What is the smallest live dataset that makes the twin useful to a real team?
Method
- 01Generate deterministic floor plans from representative organisational data.
- 02Build a navigation graph that respects rooms, doors and corridors.
- 03Ground conversational answers in the visible map and data, with a thin server boundary for secrets.
Evidence
- 01NHS Digital 3D App engineering case study.
- 02Dedicated determinism, pathfinding and movement validation harness.
- 03Production build verification recorded in the project documentation.
Next Steps
- Replace representative data with an approved operational dataset.
- Run usability testing with estates, operations and clinical stakeholders.
- Measure time to locate, understand and act on a defined operational scenario.
Related live work
Related Whitepaper records