A FHIR-native EHR shifts the workflow engine problem. Instead of orchestrating across heterogeneous data sources, the workflow engine has the FHIR resource graph as its primary data model and the FHIR REST API as its primary integration surface. The engines below have been deployed in 2026 FHIR-native EHR stacks and handle the workflow logic that turns raw FHIR resources into clinical workflows the care team actually uses. For more FHIR healthcare guides, the broader reference set covers the surrounding architecture.
Clinical Workflow Engines Used With FHIR-Native EHRs in 2026
- Camunda with FHIR adapters. The most common pick for FHIR-native deployments that want a mature BPMN-based workflow engine and have the engineering depth to wire it into the FHIR data layer.
- Temporal. Picked by FHIR-native EHR teams that prefer durable workflow execution over BPMN modeling. Fits cleanly with the FHIR REST API as the activity boundary.
- Aidbox Workflow. Built into the Aidbox FHIR-native platform, handles workflow logic against the FHIR resource graph natively without the integration step Camunda or Temporal require.
- Medplum Bots. The workflow primitive in Medplum, used by teams that want workflow logic close to the FHIR data layer with minimal additional infrastructure.
- Drools with FHIR rule sets. The rules-engine option, picked by FHIR-native deployments that need declarative clinical rules rather than imperative workflows.
The five cover the realistic workflow engine options for FHIR-native EHR teams in 2026. The right pick depends heavily on whether the workflow logic is closer to business process orchestration (Camunda, Temporal) or declarative clinical decision support (Drools, rule-based systems).
What FHIR-Native EHRs Stress About Workflow Engines
FHIR-native EHR workflow stresses three capabilities. Resource-graph-aware workflows, because most clinical workflows span multiple FHIR resources and the workflow engine has to follow the references cleanly. Long-running workflow durability, because clinical workflows often run for days or weeks (a referral cycle, a treatment plan, a follow-up sequence) and the engine has to survive deployments and restarts. And clean subscription integration, because workflow steps often trigger off changes to FHIR resources and the engine has to consume FHIR subscription deliveries reliably.
An engine that handles all three well lets the FHIR-native EHR product layer focus on the clinical workflow logic. An engine that handles workflow execution but ignores the FHIR data layer forces the team to write resource-graph traversal code that should belong to the engine.
How to Decide Which Workflow Engine Fits Your FHIR-Native EHR
The honest decision frame is whether the workflow logic in the product is closer to business process orchestration or to clinical decision support. Business-process-heavy products (referral workflows, prior authorization, care coordination) usually pick Camunda or Temporal. Decision-support-heavy products (CDS hooks, alerting, screening logic) usually pick Drools or a rules-engine alternative.
A second factor is whether the team wants the workflow engine inside the FHIR platform (Aidbox Workflow, Medplum Bots) or alongside it (Camunda, Temporal, Drools). The inside option reduces operational complexity at the cost of less flexibility; the alongside option adds operational weight but allows the workflow engine to be chosen independently of the FHIR platform.
The cornerstone EHR integration guide covers the broader patterns. The healthcare interoperability platforms roundup covers the platform layer the workflow engines sit on top of, and the EHR connectors guide covers the connector layer below.
Sources
- workflow trigger mechanism in FHIR-native EHRs - HL7 FHIR R5 Subscriptions specification
- profile baseline FHIR-native workflow engines operate against - HL7 US Core IG v8.0.0
- bulk read path some workflow engines consume - HL7 FHIR Bulk Data Access IG
