Workflows run every enterprise system — so one core engine must unify them and own every node interface: interface control is data sovereignty, the line between autonomy and lock-in.
Around the year 2000, tech giants of the day such as Microsoft and IBM drove the standardization of Web Services. In 2002, the idea evolved further into Service-Oriented Architecture (SOA) and the Business Process Execution Language BPEL4WS (later standardized as OASIS WS-BPEL 2.0). In my view, the Web Services generation marked the first time workflow engines were standardized and commercialized at scale. (WS-BPEL 2.0 specification)
Since then, workflow engines built on the SOAP/WSDL protocol stack have gradually faded from the mainstream, displaced by lighter-weight orchestration solutions, due to their heaviness and tight coupling. But the concept of the "workflow" has survived intact to this day — it lives on in BPMN, Airflow, Temporal, and the state-machine orchestration services of every major cloud provider. Its underlying logic remains the same: orchestrate a series of executable work nodes that share a common interface, so as to accomplish a specific task.
Inside an information system, a workflow typically works like this: a standard interface receives data along with runtime context, and invokes a specific executor that represents a work node. Execution may require human input or run fully automatically; once finished, the result is written back to the context area, and the next step begins until the work completed.
If a node's interface corresponds to a job position within an organization and the incoming data is a verbal instruction, the process is likely a standard operating procedure (SOP) — where the "shared interface" is the job's responsibilities and the "executor" is the person holding that position.
If the whole process takes place while building an executable from source code, it is likely a pipeline on GitHub — compiling, testing, and packaging, with every node sharing the same input-output contract.
If a node's interface receives requests from intranet servers and invokes downstream services according to predefined logic, it is likely a state-machine execution on AWS Step Functions.
In short, workflows are everywhere within enterprises. From the perspective of strengthening enterprise information management, a core workflow engine should be able to handle (or connect) all of the organization's workflows. More importantly, it must retain control over every node interface — because whoever controls the interfaces holds the power to change processes, to audit them, and to claim data sovereignty. That is the dividing line between enterprise autonomy and vendor lock-in.