Integrations

The integration burden sits with the platform, not with your other vendors.

Adapter-first is the widest moat WillowCreek digs, and its clearest commercial difference. Every external system attaches through an adapter the platform owns. No adapter is ever metered.

The adapter seam

Adapters are first-class citizens

An adapter isn't glue code bolted on after the sale. It's a first-class component with its own lifecycle, its own health, and no price tag.

Replaceable, independently

Each adapter is deployed, upgraded and rolled back on its own, without touching the platform or your other vendors' systems. Swap a technology and you swap an adapter. Your architecture stays put.

Continuously health-checked

Adapter health is watched by the platform itself and visible at the site, the estate and the provider. An adapter with no endpoint reports unconfigured, never healthy. There's no state in which silence looks like success.

Never metered

An adapter is work we do with you, not an item picked from a shelf. There's no per-adapter charge, ever. Every tenant may use every adapter at no incremental cost. Much of this market prices per adapter; we don't.

What is proven

The seam is real. Here is what runs through it today.

The claim that matters isn't a logo wall. It's that the seam works. Normalisation is proven across two independent vendor dialects in the lab: two incompatible feeds resolve to the identical canonical fix, and every consumer above the seam is untouched by the difference.

Running today

Console

The operator surfaces themselves consume the platform through the same event fabric every integration uses.

Running today

Signed outbound webhook

HMAC-signed event egress to systems you nominate. The receiver can verify origin and integrity on every message.

Running today

Facility relay

Nurse-call, fire-panel, BMS and fixed-duress class signals, carried into the platform as first-class events.

Running today

Paging protocols

TAP and ESPA 4.4.4, the protocols the installed base of hospital and facility paging actually speaks.

The adapter target catalogue

Where the adapter programme is aimed

The adapter programme targets the systems below, in a priority order set with customers. Named systems are targets, not claims of shipped connectors. An adapter is built with the customer whose site needs it, then belongs to every tenant.

Category Adapter programme targets Standing today
Positioning engines Quuppa, CenTrak, AiRISTA, Sewio, Litum, Kontakt.io and comparable BLE ecosystems Seam proven across two independent vendor dialects in the lab
Duress devices Staff badges and pendants, fixed duress buttons and pull cords, lone-worker devices Fixed-duress class signals run via the facility relay
Nurse call Austco Tacera, Ascom, Rauland Nurse-call class signals run via the facility relay; named systems are targets
Paging TAP, ESPA 4.4.4, POCSAG TAP and ESPA 4.4.4 run today; POCSAG is a target
Access control Gallagher, Inner Range, Genetec, Lenel Programme target
Video ONVIF, plus Milestone, Genetec and Avigilon Programme target
Building & I/O BACnet BMS, Modbus, IP relays BMS class signals run via the facility relay; named protocols are targets
Clinical HL7 v2.x, FHIR R4/R5, Epic, Oracle Health Programme target
Identity Active Directory / Entra ID, Okta, SCIM Programme target

The catalogue above is the adapter programme's target list. It isn't a library of ready-made connectors, and we'll tell you which is which, in writing, before you rely on any of it.

The inbound trust boundary

Every feed earns its way in

An integration seam this open needs a hard edge. Every inbound feed crosses the same boundary, and no feed is exempt.

  • Every feed authenticates. Anonymous ingest isn't a supported configuration. There's no quiet back door for a convenient device.
  • Schema validation at the edge. A malformed message is rejected at the boundary, not discovered downstream in the alarm path.
  • Per-feed rate limiting with quarantine. A misbehaving feed is throttled and quarantined on its own; it cannot crowd out the feeds that are behaving.
  • Provenance to the raw message. Every event traces back to the exact raw message that produced it: what came in, from where, and when.
  • Storm suppression. A feed going offline raises one infrastructure alarm, not one alarm per device it carried.

Outbound actions are guarded. Anything that touches physical security or life safety in an external system (a door release, a lockdown, an alarm reset) must be explicitly enabled per site, permission-scoped, confirmed and audited. A rule cannot invoke a guarded action unless it has been explicitly authorised to, and the platform never assumes authority over an external system's own safety interlocks.

The interfaces behind the seam →

How positioning feeds normalise to one canonical fix →

Bring us the systems you already own

Tell us what's on your site: engines, nurse call, paging, panels. We'll tell you honestly what runs today, and what we'd build with you, in what order.