In this leadership series hosted by Helen Dwight from BizFluency, we explore the mindsets, skills, and stories that shape impactful and future-ready leaders. In this session, Hendrik Leitner, Chief Partnerships Officer (CPO) at Otera, discusses his new book titled, Authorization: The One Decision the Enterprise Has Been Avoiding — AI, accountability, governance, and the future of work.
Helen: Welcome, Hendrik. Your book examines a topic many companies are grappling with, namely, who is responsible for authorizing, governing, and taking accountability for AI-driven autonomous processes? Tell us, what is that one decision about authorization that enterprises have been avoiding?
Hendrik: The decision is about pinpointing who exactly is responsible for authorizing, governing, and taking accountability for AI-driven autonomous processes. Traditionally, organizations coordinated people to achieve business outcomes. Today, with AI agents doing a lot of decision making as they work on automated processes, that has changed significantly. Companies now need to coordinate autonomous systems. They need to define what outcomes AI should deliver, what boundaries AI must operate within and who owns the process. Most importantly, they must define where and when human oversight is required. Enterprises are avoiding this because AI introduces a fundamentally different way of working. It is not simply another technology implementation—it changes how organizations operate.
Helen: Are companies reluctant to hand control over to AI?
Hendrik: Yes, but the issue is broader. For decades, technology was seen as something used to solve business problems, but with AI, the responsibility for decisions is shifting. Previously, IT owned technology decisions, developers then built systems, and business teams used them. Now, business process owners become responsible. It’s the finance, procurement, HR, and operations teams that must define outcomes. This means, employees with process knowledge become critical. It’s the people who understand the business, not necessarily the IT specialists, who will determine how AI should work.
Helen: Does this mean accountability is moving away from IT?
Hendrik: Exactly. AI platforms themselves are only part of the solution. The technology might represent a relatively small portion of the overall transformation. The larger challenge is bringing the organization’s intellectual property into the platform. We’re talking about business processes, decision-making logic, operational knowledge and especially experience accumulated by employees. In effect, the responsibility moves from “building software” to “designing effective processes.”
Helen: Is this more than traditional change management?
Hendrik: Absolutely. This is a fundamental shift. Companies should not simply put an “AI layer” on existing processes. They need to rethink how work gets done. New capabilities will be required. This in turn will put new demands on HR, which must help organizations educate employees, build new skills, and prepare people for AI-enabled work. Good management skills will be especially in demand in order to manage this big shift responsibly. It also means new roles and professions may emerge around redesigning and governing AI-driven organizations.
Helen: Are all AI applications the same?
Hendrik: No. There are two fundamentally different categories. The first is AI assistants. They help individuals in many ways such as writing faster, analyzing information and increasing productivity. An AI assistant is like a personal copilot. The second category includes Autonomous AI agents. These have a transformative function across the entire organization. They execute entire processes. They operate with limited human intervention and can even make decisions within defined boundaries. The book focuses primarily on this second category.
Helen: Can you give an example of autonomous AI?
Hendrik: The claim process at an insurance company is a good example. Today, multiple employees perform individual steps. People review information and decisions are made in a number of different departments. With agentic AI, the entire claims process could run autonomously. Humans intervene only where regulations require approval, for example, or, if the system flags an inconsistency. However, this requires new governance because companies must understand who is accountable when AI performs the work.
Helen: What does governance mean in an AI-driven organization? When is human oversight required?
Hendrik: Governance starts by having someone define the desired outcome. Someone needs to create operating boundaries. Checks and controls must be established. Someone needs to define the points where humans must intervene. AI agents need clear instructions from humans. The quality of the result depends heavily on how precisely humans communicate what they want. In effect, the future is not about programming AI. It’s about communicating effectively with AI.
Helen: Does this make communication a critical skill?
Hendrik: Absolutely! Working with AI is similar to having a conversation. People need to explain objectives clearly. They need to provide AI with the right context. Humans need to challenge the responses and refine instructions. The better humans communicate, the better AI performs.
Helen: What happens to organizational knowledge as experienced employees retire?
Hendrik: When an employee retires, the company risks losing valuable intellectual capital. Many organizations depend on knowledge held inside employees’ minds. People know how processes work, why and when decisions are needed. They know how problems are solved. Companies should capture this knowledge and transfer it into AI systems before it disappears. This is not about preserving old methods forever. Processes will continue to develop and younger employees will continue improving them and challenging assumptions.
Helen: Does AI replace graduates entering organizations?
Hendrik: No, not necessarily. AI can capture existing knowledge, but organizations still need people who question existing processes. New people bring in new ideas. They challenge AI outputs because they have a different perspective, and they’ll adapt systems to changing environments. The medium changes, but the need for human contribution remains.
Helen: What happens if people trust AI too much?
Hendrik: This is a significant risk. AI can provide incorrect information, even when it is not obvious. Companies and individuals must develop stronger validation processes. Instead of trusting AI blindly, people need to ask: Is this information accurate? Does it make sense? Should this decision be trusted? AI increases access to information, but human judgment remains essential.
Helen: What mistakes are companies making when adopting AI?
Hendrik: Many companies mistake agentic AI as RPA 2.0. As with robotic process automation, they’ll focus on small automation tasks and technical integration. They think adding AI onto existing processes is the way forward. Successful companies on the other hand, focus on large business processes. They’ll redesign how work is done and focus on measurable outcomes in significant areas such as procurement, claims processing and accounts payable.
Helen: What does successful AI adoption look like?
Hendrik: Companies that understand AI are beginning to create internal business hubs where teams redesign processes specifically for the agentic AI world. They’ll be shifting from where can we automate? to how should our organization operate differently with AI?
Helen: Let’s fast forward to the future. Will AI development remain concentrated in Silicon Valley?
Hendrik: The models may be developed by major technology companies, but the impact will be global. AI applications will emerge across agriculture, healthcare, aviation, manufacturing and business operations. They’ll become an integral part of our personal lives. The value will come from how different industries apply AI.
Helen: Can you give an example of AI solving real-world problems?
Hendrik: Agriculture is a strong example. AI can support farmers in a number of ways, from sensors monitoring soil to drones analyzing crops. Already, robots are performing tasks that were previously done with human labor which in turn improves the way resources are managed. Like everyone else, farmers adopt technology, because it solves practical problems and improves productivity. Technology should start with the problem, dealing with water scarcity for example, not the technology itself.
Helen: Does AI require global cooperation?
Hendrik: Of course. AI affects everyone, so governments, companies, universities, and institutions all need to collaborate. The goal is not to stop AI development, but to create informed discussion around regulation, economic impact and environmental impact. Most importantly, all these entities must cooperate to establish responsible practices for use. This is where more independent institutions and think tanks will likely emerge to help society understand AI’s impact.
Helen: What advice would you give professionals struggling with AI adoption?
Hendrik: I think companies need to understandthatthey should not adopt AI simply because everyone else is doing it. They first need to understand where AI can create genuine value. Then, it’s about developing expertise in specific areas, improving communication skills and finally, learning how to work effectively with AI. Clearly, AI will become part of almost every profession. Those people who can combine domain expertise with AI capability will become increasingly valuable.
Helen: Now, let’s talk about you. What does success look like for you?
Hendrik: For me, success would be if my book helps people to rethink how they approach AI. The goal is not to provide a fixed framework or a guaranteed solution. It is to encourage leaders to think differently, more broadly, more creatively. It’s about being less deterministic. I think the biggest impact would be if the book helps organizations ask better questions before implementing AI.
Helen: In summary, AI is not primarily a technology challenge—it is a human, organizational, and communication challenge.
Hendrik: Correct! Organizations that succeed will be those that define accountability clearly, capture their knowledge, redesign processes, and build governance structures. That all depends on learning how to communicate effectively with AI. The future belongs not to those who move fastest, but to those who understand where they are going.