ELECTRONIC AI RECORD
The system of record for changing healthcare AI deployments.
E-AIR gives healthcare AI vendors a deployment-level system of record — know what's live where, understand what a change affects, preserve the evidence behind it and reconstruct any deployment over time.
Healthcare AI doesn't stay still.
Models, prompts, configuration, feature states, integrations and workflows can differ between individual healthcare deployments.
When something changes, vendors need to understand which deployments are affected, what needs reassessing, what customers need to know and what actually became live.
AI product changes
Customer deployments are not identical
The same change can affect different deployments differently
Teams need to know what is affected and preserve what happened
A continuous record of every deployment.
Know what's live
Maintain an accurate record of the material state of every AI deployment.
Manage change
Understand deployment impact and coordinate the actions, evidence and customer responses associated with a change.
Preserve the record
Create a longitudinal history of how each deployment changes over time.
Reconstruct the past
Determine what was live at a deployment at a particular point in time.
DEPLOYMENT FINGERPRINT
Every deployment is different.
A deployment fingerprint is a structured snapshot of the variables that materially determine how an AI system is configured and behaving at one healthcare organisation at one point in time.
SHARED PRODUCT STATE
Shared across relevant deployments
LOCAL DEPLOYMENT STATE
Specific to one organisation
Deployment fingerprint
What exactly is this AI deployment doing right now?
EXAMPLE
Healthcare organisation A
SHARED PRODUCT STATE
- Model
- v2.4
- Prompt
- triage-v18
- Release
- 2026.08
LOCAL DEPLOYMENT STATE
- Paediatric pathway
- Enabled
- Chest-pain routing
- Local configuration
- EHR integration
- EMIS
- Feature set
- Configuration B
Healthcare organisation B
SHARED PRODUCT STATE
- Model
- v2.4
- Prompt
- triage-v18
- Release
- 2026.08
LOCAL DEPLOYMENT STATE
- Paediatric pathway
- Disabled
- Chest-pain routing
- Standard
- EHR integration
- SystmOne
- Feature set
- Configuration A
Same product state. Different deployment.
These fields are illustrative. The variables that materially matter differ from vendor to vendor and from product to product.
Manage the transition, not just the version.
Current state
The confirmed deployment state before the change.
Change
A model, prompt, configuration or workflow changes.
Impact
Which deployments are affected, and why.
Evidence & actions
Assessment, evidence, required actions and customer response.
Transition
The move from one state to the next.
New state
The confirmed deployment state after the change.
PRESERVED
Previous state
LIVE
Resulting state
E-AIR preserves both sides of the transition, creating an auditable history rather than overwriting the previous state.
Built around the operational reality of healthcare AI.
Deployment-level
Records what is actually live at each customer rather than only the software release.
Longitudinal
Preserves previous deployment states and the evidence behind transitions.
Vendor-first
Fits around your existing engineering, QMS, clinical-safety and customer workflows — E-AIR connects the deployment record rather than replacing the systems that create it.
Built with healthcare AI vendors.
E-AIR is being developed alongside healthcare AI vendors using real deployment and change-management workflows.
Built from inside healthcare.
Dr Richard Cumpsty
LINKEDINFounder · GP & Digital Health Clinician
Dr Richard Cumpsty is a GP and digital-health clinician. E-AIR grew from conversations with healthcare AI vendors about the practical difficulty of understanding what is deployed where, managing change across customer deployments and reconstructing deployment history.
Building or deploying healthcare AI?
We're working with healthcare AI vendors to refine E-AIR around real deployment and change workflows.