Robotic Process Automation (RPA) is the deployment of software robots to execute a task which would have been manually executed by humans. Its central aim is to reduce the direct involvement of humans in the execution of a process.
RPA is not a replacement for humans. For one, no robot (nor machine) can carry out cognitive-based tasks. And such task (or activity) is a fixture in any organisation. Humans will always be needed in the scheme of things.
RPA is best suited to repetitive activities (e.g., filling out static forms with similar info all the time, etc). Studies have shown humans find this kind of tasks to be mundane, laborious and error-prone. Getting humans to complete them will almost always result in reduced productivity for the organisation (due to errors made, hours lost correcting such errors, dips in execution pace, staff demotivation, etc).
An ideal deployment of RPA is one where the software robots work in conjunction with humans. Here, the robots act as assistants, completing the laborious repetitive tasks as necessary. This frees up humans to ably handle the cognitive ones. Such collaboration drives up productivity and reduces cost (due to hours saved, among other things). To this end, the software robots are usually aptly named “digital assistants”.