Robotic process automation (RPA) is the use of software with artificial intelligence (AI) and machine learning capabilities to handle high-volume, repeatable tasks that previously required only humans to perform. In short, there is at least a problem with traditional Business Process Management (BPM) systems, as they cannot suggest the best combination of tasks, people and timings, which can increase the benefits of running them, while reducing the costs and risk factors. Yet, it is an irrefutable fact that the current business environment is highly dynamic. On the one hand, we need to be more efficient to execute what is operational and obvious, releasing scarce resources for more critical areas. Then, dealing with business process management and automation, a common claimed benefit is associated with the improvement of performance. In addition to this and other potential benefits, we also highlight some potential operational risks from the adoption of AI-based systems like RPA. The acceleration in the business context makes it more difficult to predict what changes will occur and how they can affect the technological solutions used in the increasingly automated business processes. We point out the fact that immature or not well-trained models can eventually decrease productivity and increase errors from unsupported or even wrong decisions. We present a case study in the banking sector, which illustrates some examples of benefits and risks arising from BPM solutions that use AI-related agents/artifacts.