Organization type · Health Systems
Enterprise deployment for multi-site health systems, with central governance, local protocol control, estate-wide routing and a sovereign tenancy that can scale without replacing the systems already in place.
In the product
Enterprise console, estate overview
LiveTenancy: eu-west-sovereign-02 · 11 sites · 34 services live
Policy: system protocol v14 · 4 site overrides, all within permitted scope
Primum AI
Systems connected: Epic (6 sites) · Meditech (3) · athenahealth (2) · legacy PMS (1, shadow user)
Today: 41,208 contacts · 68% resolved without staff · 1,914 escalations routed
Availability 99.99% · p95 response 42ms · 0 data egress events
Primum AI
Rollout: nephrology service line staged for Monday, inheriting 4 existing connections
Tenancy
One tenancy · one audit trail · one report across the estate
Overview
Two EHRs from an acquisition that never finished, a separate scheduling system in radiology, a legacy PMS at the sites that joined last year, and a contact centre stitched across all of it. Every improvement proposal starts with a migration, so nothing gets improved and the phone estate stays where it was in 2014.
Primum AI sits above that estate instead of replacing it. OIL, the Operational Intelligence Layer, reaches each system the way that system allows: FHIR and HL7 where the interfaces exist, and by driving the session as a shadow user where they do not. Policy is set centrally and inherited by every site, which can then vary what it's permitted to vary. Nothing is migrated, and the first service is live in a day.
Capabilities
The difficult part is keeping governance consistent while allowing sites to operate differently where they need to.
Protocols, escalation rules and scripts are defined once at system level and inherited downward. A site or a service can override only what governance has marked as variable, and every override is recorded and reviewable.
Epic, Cerner, Meditech, athenahealth and whatever the last acquisition brought with it. Read the chart, reason over it, write the result back, over standards where they exist and as a shadow user where they do not.
A patient calling one number reaches the right site, the right service and the right skill, with capacity balanced across the system instead of trapped inside whichever site answered.
Dedicated infrastructure in the jurisdiction you nominate, with no data egress, no shared model training and an audit trail that satisfies both the regulator and the board.
How it works
Most deployments begin with one service, prove the workflow, and then extend to more sites and specialties using the same layer.
01
A dedicated sovereign environment in your jurisdiction, with identity, retention and residency policy configured before any patient data exists in it.
Isolated infrastructure
02
Each clinical system reached over the interface it supports, with the connections held on the layer instead of rebuilt per service.
FHIR / HL7 / shadow user
03
System-level protocols, escalation rules and permitted variation set once, then inherited by every site and service that comes online.
Inheritance model
04
Service by service, site by site, with volume, outcome, deflection and cost reporting consolidated across the whole estate from the first day.
Estate reporting
01
A dedicated sovereign environment in your jurisdiction, with identity, retention and residency policy configured before any patient data exists in it.
Isolated infrastructure
02
Each clinical system reached over the interface it supports, with the connections held on the layer instead of rebuilt per service.
FHIR / HL7 / shadow user
03
System-level protocols, escalation rules and permitted variation set once, then inherited by every site and service that comes online.
Inheritance model
04
Service by service, site by site, with volume, outcome, deflection and cost reporting consolidated across the whole estate from the first day.
Estate reporting
Benefits
The connections and governance created for the first service can be reused by the next ones.
A new service inherits every connection the layer already holds. What used to be an integration project becomes a configuration change.
Central protocol inheritance means the same presentation gets the same triage whichever number a patient rang.
Volume, acuity, deflection, escalation, cost per resolved contact and outcome, consolidated across every site instead of reconciled from four contact centre exports.
In practice
All on the same tenancy, all inheriting the same governance.
Estate-wide routing by service and capacity, replacing four separate switchboards without touching any of the underlying systems.
Sites on Epic and sites on Meditech run identical triage and booking flows, with write-back handled differently underneath and identically above.
Out-of-hours coverage consolidated across the estate onto one on-call structure, with site-specific escalation preserved.
Cardiology, oncology, orthopedics and six more brought live in sequence, each inheriting the connections the first one paid for.
Not the flagship, the one running the system nobody wants to touch. We'll connect it and run a live service there, and the rest of the estate is easier than that one.