I’ve been working with UiPath for years. Now everyone is talking about "Generative AI" workflows. Does this mean traditional RPA is dead, or do these two technologies work together to create a more robust automation system?
3 answers
RPA isn't dead; it's getting a brain transplant. Traditional RPA is "rule-based"—if X happens, do Y. It struggles when the data is unstructured (like a long email or a messy document). That’s where AI comes in. The most effective workflows I’ve seen lately use RPA to handle the "clicks" (logging into a legacy system) and AI to handle the "decisions" (interpreting what a customer actually wants). We call this "Intelligent Automation." Instead of replacing your UiPath skills, you should learn how to call an AI API from within your RPA workflow.
Do you think the cost of adding LLM calls to every RPA transaction makes it too expensive for high-volume tasks? I’m worried about the 'hidden' costs of tokens in these automated loops.
RPA handles the hands, AI handles the eyes and brain. You need both to fully automate any complex business process in a modern office.
That’s a great analogy, Gary. I've started telling my clients that RPA is the muscle and AI is the intelligence. You can't run a warehouse with just one or the other.
Wayne, cost management is definitely the new challenge. The trick is to only use the AI when the RPA fails. You write a script that tries to process a form using standard rules first. Only if the "confidence score" is low do you trigger the LLM to interpret the data. This "hybrid" approach keeps the costs down while significantly increasing the automation's success rate.