Measuring the Autonomous Intelligence of Your Process Via Freedom Levels
As processes become more intelligent, we will likely want to measure the level of intelligence, as I discussed in "How Smart Is your Business?" Doing this will give organizations an idea where they are in a continuum of trying to become a smarter business over time. This month I cover the second "A" portion of the "ISAA" [Intelligent, Social, Agile, Autonomous] framework described in that column.
I propose the following five levels of process autonomy that build on each other.
1. Programmed Behavior
A process can be completely prescribed with some levels of agility, but the permutations and combinations are preplanned. This way control is exercised by the process managers and operators, and there is a high dependence on a command and control approach. This is very proactive approach — but quite rigid.
2. Permitted Actions
A process can suggest alternative actions for process managers or ask permission to act in a way that was not expected. This requires a level of intelligence to point out emerging patterns and to suggest proper responses. Nothing happens without some level of permission.
3. Act First, Then Notify
A process can watch and learn and take action based on some proactive anticipation. In this case, the process manager is notified in a timely manner of the process's decision and action. A process manager can then make appropriate actions, if the process is wrong, or can lower the level of freedom for this kind of process instance or case.
4. Act with Constraints and Goals
A process on its own can be goal seeking in nature and call in the proper analytics to self-adjust goals to reach optimum outcomes based on static or dynamic weightings of goals. Processes (or appropriate process snippets) can be kept away from out-of-bounds conditions through constraints. This creates a balance between freedom and negative boundary conditions.
5. Interactive Independent Action
Processes or process snippets can interact with other process snippets or the Internet of Things to create a dynamic response that requires automated collaboration of automation and measures. These snippets can systematically flock temporarily or permanently to deal with emerging patterns.
There are definite levels of autonomous intelligence that processes can exercise. We will need learn to utilize various levels of autonomy over the coming years.
# # #