Daniel, The persistence is fully pluggable. While our JPA implementation generates one row in the database for each process instance, you can plugin in your own implementation that write the data you need somewhere else, for example inside one of your domain objects. We provide marshalling / unmarshalling capabilities that can generate a binary blob representing the process instance state so that it could easily be stored as part of your model. For example, the following junit tests uses a simple Map implementation: https://github.com/krisv/jbpm/blob/master/jbpm-persistence-jpa/src/test/j... Regarding getting existing processes in jBPM without actually executing them, yes, also that is possible, by creating your own process instance type and simply generating the process instance objects whenever necessary. We used this approach for example to aggregate BPMN2 and WS-BPEL processes together, where the WS-BPEL processes were not really being executed by the jBPM engine but by the JBoss RiftSaw engine. Note that if you want to combine external process instances for monitoring / management, you might want to consider doing the integration on the process history log (meaning both engines write out to the same history log in the same format), which is easier but still allows you to use one management / monitoring tool across process instances. I hope this makes sense ;) Kris Daniel Migowski wrote: