Between Shell Script and Self-Deception

What is automation really? The basic idea of automation is simple: tasks or processes are designed in such a way that no human intervention is required and they are more precise, faster and error-free at the same time. In practice, this means that scripts, workflows, tools or bots take over defined activities such as rolling out software, providing cloud resources or sending invoices.

It is important to note that automation is not a technology, but a way of thinking. It can be done manually using shell scripts, visual workflow tools or robotic process automation (RPA), in which a bot operates user interfaces. The much-cited term "hyperautomation" ultimately just means: automate everything that can be automated. Whether this always makes sense is another question.

Where automation actually helps

The benefits are undisputed if automation is used correctly. It reduces human error in routine tasks, accelerates workflows, enables scalable processes and brings order to IT landscapes. In modern DevOps environments, continuous integration and delivery are unthinkable without automation. It has also become indispensable in infrastructure management, for example when provisioning servers. Standardization is an often overlooked advantage: automated processes always run in the same way, are ideally self-documenting and reliable. This is particularly beneficial in security-critical areas.

Limits of automation

Nevertheless, caution is advised, because automation scales everything, including errors. An incorrectly configured deployment runs reliably in an automated process - and it goes wrong. And if you automate bad processes, you don't get better processes - you get badly automated ones. A common misconception is that automation is cost-saving per se. In fact, the initial outlay is high. Processes need to be understood, rethought, modeled and tested.

Maintaining automated systems is more complex than many white papers suggest. If you don't think long-term here, you create a new kind of technical debt. Not all automation is ethically sound either. Today, decisions about people - for example in application processes or when granting loans - are sometimes made automatically.

But what happens when the algorithm discriminates? When cause and effect can no longer be traced? Automation also needs a sense of responsibility. Another mistake is to equate automation and autonomy. A script that makes a backup on Sundays is automated. A system that recognizes when a backup is necessary acts autonomously - and that is a completely different level of complexity.

Automation versus autonomy

These terms are currently blurred in the marketing fog surrounding artificial intelligence. However, autonomy requires contextual understanding, the ability to learn and situational awareness - skills that machines only possess to a very limited extent.

The human being remains decisive

Despite all the advances, automation needs people to design, maintain and scrutinize it. Automation takes away work - but only if we have defined it. Without a deep understanding of the underlying processes, automation becomes a black box. Therefore, not everything that can be automated should be automated. And not every human decision can be replaced by rules or data.

Continue to the partner entry: