Guy van Wert (Principal Product Manager, UiPath) took Hyper ‘25 on a deep dive into the next evolution of Robotic Process Automation (RPA). He demonstrated how an API layer makes RPA more resilient and introduced agentic automation: intelligent AI agents that make autonomous decisions. "More than just a bot, a digital employee that detects and solves problems without constant human intervention."
UiPath started as an RPA platform, helping organizations automate repetitive tasks without the need for complex software development. By mimicking human interactions with systems—such as opening a webpage, logging in, and retrieving information—companies could save time and reduce costs.
However, Van Wert emphasized that RPA has its limitations, especially when systems and interfaces change, causing bots to stop functioning properly. “RPA is great, but if an interface changes, your automation breaks. We need a smarter approach.”
To make businesses less dependent on fragile user interfaces, UiPath added an API layer to its platform. The necessary iPaaS technology was acquired through the Cloud Elements acquisition.
Instead of bots reading screens and responding, systems now communicate directly via APIs. This makes automations more resilient, reduces dependence on visual interfaces, and lowers maintenance costs while making processes more scalable. “We wanted to bridge the gap between RPA and API automation so that companies can remain flexible.”
Next step: Agentic Automation
The core topic of Van Wert’s presentation was agentic automation, a new approach where intelligent software agents play an active role in automation processes. Unlike traditional RPA bots, which execute tasks from start to finish, these agents work interactively and adaptively. Rather than following pre-programmed tasks, agents can make autonomous decisions and respond to complex situations. This means that workflows no longer have to be fully linear—an agent can wait for input, independently collect data, and even communicate with a human employee if necessary.
“An agent is not just an advanced bot. It is a digital co-worker that can recognize and solve problems without constant human intervention.”
We’re not taking humans out of the loop, but we make sure they only step in when it truly matters. This significantly speeds up the process
Guy van Wert, Principal Product Manager UiPath
Use case: automated invoice processing
A concrete example Van Wert provided was invoice processing. Traditionally, an RPA bot reads an invoice from a shared inbox and attempts to process it in an ERP system such as SAP or NetSuite. However, if a purchase order number is missing or a supplier is unrecognized, the process gets stuck, requiring manual intervention. With agentic automation, this process becomes much smarter. An AI agent can independently search for missing data, analyze previous transactions or contact the supplier to obtain additional information.
“We’re not removing humans from the loop, but we ensure they only step in when truly necessary. This significantly speeds up the process and reduces the risk of errors.”
Keeping your Robots under control
Another key takeaway from the presentation was agentic orchestration, a feature that helps businesses maintain oversight of their automated processes. Many companies run hundreds of RPA bots, making it difficult to manage them effectively. Agentic orchestration adds transparency to automation, says Van Wert. “Instead of a chain of bots triggering each other without clear oversight, we provide companies with a visual representation of how everything works together.”
The Switzerland of AI
Agents are AI-driven, but Van Wert emphasized that UiPath wants an open AI ecosystem. Instead of imposing a single AI model, the platform allows businesses to choose their own AI models, such as OpenAI’s GPT, Google Gemini, or Anthropic Claude. “We want to be the Switzerland of the AI world—neutral. Companies should be able to decide for themselves which AI works best for their needs.” This means that organizations can leverage the power of AI without being locked into a single vendor or platform.
One of the main challenges in AI-driven automation is control and reliability. AI models can make mistakes or even "hallucinate". UiPath addresses this by developing a framework where AI agents only have restricted access to systems and can execute only specific tasks. Organizations can determine their own level of automation, such as having AI agents only make suggestions, while a human makes the final decision. Alternatively, AI agents can be fully autonomous in specific scenarios. “This hybrid approach ensures that businesses benefit from AI automation without the risk of uncontrolled AI decisions.”
Conclusion
According to Van Wert, the introduction of agentic automation marks a fundamental shift in how companies approach automation. Instead of static bots that complete tasks from A to Z, UiPath introduces intelligent agents that work adaptively and allow for human interaction when needed.
With these new tools, UiPath offers a scalable solution for automating complex business processes. “The future lies in agent automation and the intelligent collaboration between humans and machines.”
Do you want to know more?
Edwin Nooijen
Sales Director
