Launching the robot to production 🚀
You have developed and tested your robot, and it works great! Time to put it "in production" and let it run free.
Before you can put your robot into production, we recommend following a checklist like this one:
Is the code correctly versioned and available for another developer if you are not available to work on the robot in the future?
Does the implementation follow naming best practices?
How will the robot be executed? Is it supposed to run on a schedule or an API call from another system?
Have you set up notifications for failed runs? It is good to let someone know if the robot is having a hard time.
How will you monitor the robot after launch? Monitoring the robot is essential. This way, you can promptly react to possible issues and ensure the automation produces the expected results.
In addition to detecting failures, how do you verify that the robot is actually doing the right things and producing the expected outputs and end states? It is not enough to know that "the robot is running" if it's completing tasks incorrectly or producing incorrect results.
Who will react if you need a prompt resolution to an issue in production (both from the technical and the business side)? Internal staff, or maybe an external contractor?
Is it possible to safely run a failed process again, even fully or partially? If not, prepare a backup plan for handling those scenarios.
What affected parties need to be notified if there is an issue? Imagine, for example, a failure in an automated payroll process: that would affect many people in the organization.
Is there internal or external expertise and knowledge to handle production issues?
Is the original implementation team active and available? If not, is there internal or external staff available to handle issues or maintenance? Is the documentation good enough for someone else to maintain or fix the implementation or remedy production issues?
Are the original business stakeholders with the required process knowledge available? If not, is the process documented well enough so that someone else can fully understand it?
As you can see, launching automation to production does not mean you can then forget about it completely.
Once active, the automation is essentially an employee, or multiple employees, completing essential tasks in a production environment. Proper error handling and issue propagation is important. Monitoring is critical. Is the automation running? Is it doing the right things? What to do when eventually something needs to be done to update the automation or resolve an issue?