Thank you again, Steve. We need to make all kind of changes to the rules, however, small changes are more frequent than significant ones. So, every time we are deploying a new rule, there is a risk of it either not compiling or failing properly follow business logic. We can trust users with any changes, however, moving the code to production is a big deal. This should be vetted by an authority figure, and there must be a simple and transparent rollback plan. Yes, we want to be very risk-averse. Ideally, we would like to have: 1. A staging environment where automation tests are run; 2. A change can be deployed to production only if all automation tests have passed; 3. Some kind of administration console from which a change can be manually deployed to production (via uploading to production Maven repository or in some other way); 4. Production Drools system picking up new changes without interruption of service; 5. Production console function allowing a one-click rollback of a recent change; Alex P.S. Sorry, it looks like I have not mastered proper replying to a forum thread.