Let’s Talk: Agentic AI

The rise of Agentic AI holds the promise to revolutionize industries and reshape the way humans and machines work together. At its essence, Agentic AI consists of systems capable of operating independently — making decisions, pursuing objectives, and taking action with little to no human oversight. Unlike traditional AI, which often just responds to commands or processes data in the background, agentic AI can perceive its environment and make informed decisions on its own. Use cases are manifold – from autonomous vehicles to personal assistants and customer service automation to healthcare and more. We asked Sabrina Blaukopf, Director New Technologies at Papyrus Software to share her thoughts on Agentic AI.

EDITORIAL: Papyrus Software has established itself as a leader in delivering innovative solutions for business users over many years. Are there any plans for Papyrus Software to incorporate Agentic AI into its future software developments?

Sabrina Blaukopf: Yes, in fact, our latest no-code innovation for business experts, Papyrus Converse, already leverages large language models and Agentic AI. Everything we have in place here not only empowers business users but also enables AI agents and AI capabilities within our process applications.

We’ve designed Converse specifically for businesspeople to build, deploy and deliver complex processes with ease. You know, most of the customer-centric workflows, collaborative workflows, as well as decision-heavy processes where different decisions can lead you in completely different require a certain level of flexibility, and this is precisely Converse’s forte.

Here we use a smart, friendly AI agent, whom we affectionately call Owlfie. Owlfie is built on technologies like Papyrus Machine Learning and IBM watsonx GenAI, and designed to help business users define and optimize their processes from the very beginning. It remembers everything it learns along the way and leverages this know-how to understand the unique context of each case or value stream during business process execution. It’s like having a trusted advisor right there helping you make the right decision and follow best working practices in every step.

You don’t need to do anything special – just work as you normally do. Much like the human brain, the Owlfie’s knowledge grows with its experience as time goes on. We take it a step further by incorporating insights from all customer interactions—calls, emails, and documents—seamlessly feeding into Owlfie’s business knowledge and your processes, helping guide and support your employees. Owlfie delivers insights straight to the employees – whether they’re exploring product features, troubleshooting technical issues, or seeking recommendations, they can simply ‘converse with their data’.

Owlfie navigates documents, decodes customer inquiries, crafts responses, and generates reports, adding a personal touch to your customer relationship depending on the overall sentiment of the conversation.

EDITORIAL: What do you believe is Owlfie’s most important capability?

Sabrina Blaukopf: I would say its greatest strength is the ability to quickly extract relevant information from vast amounts of data in just a matter of seconds. We recently had an interesting use case where Owlfie helped summarize the information from multiple committee documents and pinpoint exactly where specific details were found, making it easy for anyone to look them up. This is a very important feature.

Owlfie can also suggest the next steps in your workflow, based on what it has learned, or handle tasks on your behalf. For instance, it can create a claimant in a claims process, search for missing information across various databases, and input all relevant data into the claim case. It can also autonomously send emails from the same interface and read the response from the customer. All interactions are in plain business language in a chat style, making the entire process really simple and intuitive.

owlfie.ai

EDITORIAL: The potential of agentic AI is exciting, but it also brings important concerns about safety and control. What are your thoughts on this?

Sabrina Blaukopf: I agree – making sure agentic systems behave reliably is absolutely essential. That’s why we have built extra steps to make sure that Owlfie clearly understood what you asked. We intentionally do this to ensure safety. Whenever you interact with the agent, it will always provide feedback to show what it understood from your input. This is all the more important as the whole communication with Owlfie is done in natural language, and language is not specific –  it’s ambiguous. Therefore, it’s important that we make sure that what was understood is correct.

In addition, Owlfie has to follow the same business rules as your employees. It can’t perform certain actions without permission. In addition, any changes that impact the system always require final approval from the business user, ensuring control and oversight.

EDITORIAL: What comes next?

Sabrina Blaukopf: We are currently working with IBM to make use of IBM Orchestrate, where we can extend our current agentic capabilities to a multi-agentic system. Here, you can employ a range of specialized agents—each responsible for specific functions such as data collection, decision-making, or task execution. These agents can effectively process large amounts of information and use tools to carry out actions. They can reflect on their performance, and adjust if needed, working in an orchestrated manner to achieve shared goals.

Here, too, we have established clear safeguards.

We keep abstracted these additional agents that you might have for different tasks away from the business user so that the business user interacts only with one agent – Owlfie.

Also, regardless of how many agents are working behind the scenes, orchestration and control come always from our platform. This makes sure that agents that work for the business user cannot access data or perform actions that the user isn’t authorized to see or do. So, we use our infrastructure for our business users and for the agents to make sure that this orchestration and facilitation happen in the right way.

EDITORIAL: Your final thoughts on Agentic AI?

Sabrina Blaukopf: The journey ahead is filled with opportunities. We’re excited to collaborate with companies that, like us, believe in approaching each step carefully and intentionally, and creating something truly remarkable along the way.

Let’s Talk! Connect with us to continue this conversation on Agentic AI!