Podcast Summary: The Agile Brand with Greg Kihlström®
Episode #686: Autonomous Innovation Using Predictable AI Agents with Peter van der Putten, Pega
Release Date: June 6, 2025

Introduction

In Episode #686 of The Agile Brand with Greg Kihlström®, host Greg Kihlström engages in an insightful conversation with Peter van der Putten, Director of the AI Lab and Lead Scientist at Pega. The episode delves into the evolving landscape of artificial intelligence (AI) in enterprise settings, focusing on the concept of predictable AI agents and their role in fostering autonomous innovation within businesses.

Understanding Predictable AI Agents

Peter van der Putten introduces the concept of predictable AI agents, distinguishing them from the more commonly discussed generative AI models like ChatGPT. He emphasizes that while generative AI is impressive in its capabilities, it remains largely passive, responding to prompts without autonomous decision-making.

“Generative AI models... are amazing what they can do, but they're also quite passive. When you think about agentic AI, we want to have AI systems that can operate in a world where they can sense the environment, understand goals, plan, and take actions to achieve specific outcomes.” (04:28)

Agentic AI represents a shift towards more active AI systems capable of autonomous decision-making, planning, and utilizing various tools to achieve defined objectives. This evolution is critical for enterprises aiming to harness AI for innovation without compromising control and predictability.

Agentic AI vs. Generative AI

The discussion highlights the limitations of generative AI in enterprise applications, particularly concerning predictability and control. Peter underscores the necessity of integrating governance mechanisms to ensure AI agents operate within defined boundaries, especially in sensitive and regulated industries.

“Predictable AI agents... address concerns that enterprises have around how can we empower these agents so that they have the right tools, but also govern them so that they are doing the right things.” (05:58)

This balance ensures that while AI agents can innovate and streamline processes, they do so without overstepping into areas requiring human oversight or adherence to strict regulatory standards.

Practical Applications in Regulated Industries

Peter provides concrete examples of how predictable AI agents can be deployed in regulated sectors such as insurance and finance. He discusses the application of AI in insurance claims processing, where agents can autonomously handle routine tasks while complex decisions—like medical assessments—remain under human control.

“In a life insurance policy underwriting process, an agent might determine whether a medical check is necessary. If so, governed business rules would trigger the appropriate process, ensuring compliance and accuracy.” (07:40)

This approach not only enhances efficiency but also maintains the integrity and compliance necessary in regulated environments.

Guardrails and Governance in AI Deployment

A significant portion of the conversation centers on the importance of establishing guardrails to govern AI behavior. Peter emphasizes that without these safeguards, AI agents could produce inconsistent or erroneous outcomes, undermining trust and reliability.

“Without guardrails, you're kind of just hoping. With a system like this, you're using the right tool at the right time and only within those parameters.” (21:26)

He draws a parallel to human employees, asserting that AI agents should be treated as responsible actors within the system, with their actions audited and tracked meticulously to ensure accountability.

Client Feedback and Real-World Implementations

Peter shares early client feedback and case studies showcasing the effectiveness of predictable AI agents. Notably, he references Rabobank, which has implemented agentic AI in areas like financial economic crime analysis. This implementation involves agents handling vast amounts of unstructured data to identify potential fraud, demonstrating the practical benefits and challenges of deploying AI in complex scenarios.

“Rabobank... started with a knowledge agent to answer questions based on work instructions and expanded to building a full client profile to enhance their economic crime detection capabilities.” (24:09)

These examples underscore the versatility and potential of AI agents when appropriately governed and integrated into existing workflows.

Demystifying AI and Future Outlook

Peter aims to demystify AI, advocating for a balanced perspective that recognizes both its potential and limitations. He encourages enterprises to engage hands-on with AI technologies to better understand their functionalities and applicability.

“Stay curious and get real. Go beyond just the hype and play around with the technology. Getting hands-on is a good way to do a sanity check on what's behind it.” (27:19)

Looking ahead, Peter envisions a future where predictable AI agents are integral to enterprise operations, driving innovation while adhering to essential governance frameworks. This evolution is pivotal for businesses striving to remain agile and competitive in an increasingly AI-driven world.

Conclusion and Key Takeaways

Episode #686 provides a comprehensive exploration of predictable AI agents and their transformative potential in enterprise settings. Key takeaways include:

• Agentic AI represents a proactive shift from passive generative models, enabling autonomous decision-making and innovation.

• Governance and guardrails are essential to ensure AI agents operate within defined parameters, maintaining predictability and compliance.

• Real-world implementations, such as those by Rabobank, demonstrate the practical benefits and considerations of deploying AI in regulated industries.

• Demystifying AI through hands-on engagement and continuous learning is crucial for businesses to effectively integrate AI into their strategies.

Peter van der Putten’s insights offer valuable guidance for marketing and technology leaders aiming to leverage AI responsibly and effectively, ensuring long-term business value and customer satisfaction.

Learn More:
To delve deeper into the topics discussed in this episode, visit www.agilebrandguide.com or follow Peter van der Putten and Pega for ongoing updates in AI innovation and application.