For years, RPA (Robotic Process Automation) has been a reliable way to automate structured, repetitive tasks. But with the rise of Agentic AI, a new chapter is dawning.
To explore that shift, Marijn van de Poel, CPTO (Proposition & Technology) at Ciphix, sat down with two colleagues at the forefront of this development: Jon and David. Jon is the UiPath Tech Lead at Ciphix, and is also one of the two UiPath ‘Most Valuable Players’ (MVPs) in the Netherlands. David is our UiPath Practice Leads.

In this interview, they discuss what RPA continues to do well, where its limitations lie and how AI is helping to build workflows that are smarter and more autonomous without losing control.

 

Marijn: Looking back at the impact of RPA and where we are today, where do you think the strengths of RPA lie, and where are the limitations?

Jon: RPA excels at clear, repetitive tasks, think data processing. But as soon as data formats or screens change, it starts to falter. People are better at that: they can adapt.
The ideal situation? Combining the reliability of RPA with the flexibility of AI but without giving AI blind control.

 

Marijn: Many people see RPA primarily as a tool for automating single tasks. How do you see that and what changes when you add AI?

Jon: RPA can certainly automate entire processes, but AI makes it easier to break those processes down into smaller, modular workflows. That makes them more manageable and less prone to error. AI gives RPA the ability to make choices within such a flow. That’s where autonomy comes in.

 

Marijn: Many organizations still rely heavily on RPA. How do you see the transition toward AI-driven solutions?

Jon: Existing RPA solutions remain valuable for structured tasks. What is changing is the way they are built: faster, thanks to AI support. And maintenance becomes easier. That hybrid approach increases efficiency AND relieves the burden on RPA teams.

 

Marijn: You also sometimes hear that RPA is “outdated.” How do you view that, especially in view of companies that are still working with it in abundance?

David: That RPA would be “dead” is really an exaggeration. Technology is changing: think prompt engineering and smart workflow validation. RPA is not disappearing, it is evolving. And those who recognize that will stay ahead.

 

Marijn: Can you briefly explain what Agentic AI means to you?

Jon: Agentic AI is AI that is given a specific purpose within a business process. You give it the right tools, input and output structures, and within those frameworks it is allowed to act independently.

 

Marijn: How did RPA actually pave the way for Agentic AI?

Jon: RPA has taught us to think in modular, testable pieces. That forms the basis on which you can orchestrate AI within business processes and thus AI also contributes to decision-making and process control.

 

Marijn: So the transition is all about giving ‘robots’ more autonomy to adapt processes?

David: Exactly. You want to move to a form of RPA with more self-initiative, closer to how people think and work, but controlled.

 

Marijn: Can you give a concrete example of how AI enhances traditional automation?

Jon: A good example is address processing. A standard RPA script stumbles over variations or messy input. Add AI that can analyze unstructured addresses and correct them with Google Maps, for example, and you get a much more accurate as well as smarter process.

 

Marijn: Agentic AI is promising, but ultimately a means and not an end. How do you think companies should start with use cases?

David: Start not with the technology, but with the process. Outline that, break it down into steps and choose the right technology for each step: RPA, AI or an integration. Make sure you can remain flexible.

 

Marijn: Technology is changing at lightning speed. How do you make sure your solutions stay relevant?

Jon: Focus on the business problem, not the tech stack. If the goal is clear, the technology can continue to evolve around it.

 

Marijn: Finally, do you run the risk of forgetting the power of traditional automation because of all the AI hype?

David: Sure. The hype creates attention, but it also creates confusion. You have to keep a sharp focus: sometimes a simple rule-based solution is just the best choice. It’s all about balance and the right technology for the right challenge.

 

Marijn: Finally, what do you want organizations to take away from this transition from RPA to Agentic AI?

Jon: Make sure you have a plan. Build modular, testable, and with a good understanding of your processes. And don’t forget process intelligence; that’s essential to monitor and keep improving.

 

Thanks to Jon & David for this conversation. Wondering how your organization can get started with Agentic AI? We’d love to think with you about a first step that does work: concrete, scalable and aligned with your processes.