Häufig gestellte Fragen
What is RPA (Robotic Process Automation)?
RPA (Robotic Process Automation) refers to the use of software robots, so-called bots, that carry out rule-based and repetitive tasks in digital applications by mimicking human interactions with user interfaces. Instead of transferring data manually from one system to the next, a configured bot performs these steps automatically and according to fixed rules. A defining characteristic is that RPA operates at the presentation layer and does not modify the existing applications themselves. In the ERP environment, RPA frequently closes the gap between systems for which no programmatic interface is available or economically feasible.
When is RPA worthwhile?
RPA is worthwhile above all for high-volume, stable and clearly rule-based processes without frequent exceptions, for example transferring order data from supplier portals, creating master data or reconciling documents. Suitable processes can often be automated within weeks, because RPA works via the user interface and does not intervene in the systems' business logic. Fluctuating or rarely executed processes, by contrast, seldom justify the maintenance effort. To identify suitable candidates, process mining is often used upstream, making workflows visible on the basis of system logs and uncovering automation potential.
What does RPA cost?
Cloud RPA is frequently licensed per bot and month; for an unattended robot from established vendors such as UiPath or Automation Anywhere, depending on the plan, it falls in the range of a few hundred euros per robot, while Microsoft Power Automate with user-based subscriptions starts significantly cheaper at around 15 euros per user per month. Experience shows, however, that licence fees account for only a smaller share of total costs; the larger share goes to process analysis, implementation, testing, training and, above all, the ongoing maintenance of the bots. Because RPA works via the user interface, bots can break during software updates or interface changes and must be repaired, which permanently increases operating costs. A reliable cost calculation should therefore factor in these ongoing efforts from the outset.
What is the difference between attended and unattended RPA bots?
Attended bots run at the workstation and are deliberately triggered by an employee, for instance as an assistant that takes over individual steps during a transaction or processes data on demand. Unattended bots, by contrast, work unsupervised in the background, triggered on a schedule or by defined events, and are suited to high-volume back-office tasks without human intervention. Both variants are usually managed by an orchestrator that distributes, schedules, monitors and logs the bots. Many companies combine both models in order to relieve employees directly as well as automate entire processes end to end.
How does RPA differ from an API integration?
RPA operates at the user-interface level and mimics human input, whereas an API integration exchanges data directly at the technical interface level between systems. An API-based connection is generally considered more robust, faster and lower-maintenance, because it works independently of layout changes to the interface. RPA is more fragile by comparison, since a bot tied to the interface structure can fail after an update and must be repaired. In practice, RPA is therefore mostly understood as a pragmatic transitional or supplementary solution when no stable interface is available or providing one would be too costly, not as a strategic architecture.
Which RPA vendors exist and how big is the market?
Tools that are widespread on the market include UiPath, Microsoft Power Automate, Automation Anywhere and SS&C Blue Prism; UiPath is regarded as the market leader and, in industry analyses such as the Gartner Magic Quadrant, is counted together with Automation Anywhere among the leading pure-play RPA vendors. Market analyses estimate the global RPA market volume very differently depending on scope, with figures of around nine to over thirty billion US dollars for the middle of the decade and consistently double-digit annual growth rates. Which solution fits depends on the system landscape, existing interfaces, licensing model and internal expertise, not on a blanket ranking. Increasingly, classic RPA is merging with artificial intelligence into so-called intelligent automation or hyperautomation in order to incorporate unstructured data and decision logic as well.