AI transformation: Who is responsible for what as AI scales across the organization?
A division of responsibilities that works
In our experience, organizations make the most progress when three areas of responsibility are clearly assigned:
- Business leaders and process owners shape the processes from a business perspective. They design processes around the desired outcome, prioritize which ones to redesign first and make employees available to work on them.
- IT leaders provide the technical foundation: architecture, data, integration with core systems and security.
- An AI governance body defines the guardrails within which teams can experiment independently, without seeking approval for every individual idea.
Alongside these responsibilities, new roles and bodies are emerging:
- A person responsible for AI transformation, whether as Head of AI, AI Transformation Lead or from a strategy role. The title matters less than having the mandate to set priorities across organizational boundaries.
- AI leads within business areas, who identify use cases, contribute subject-matter expertise and drive implementation in their respective areas.
- A delivery unit, often called an AI hub, AI task force or AI factory, which brings together technology, data and methodological expertise. It works with business areas to make processes AI-ready and build prototypes. It is often deliberately placed outside IT so it does not get lost in day-to-day operations. In that case, the division of work with IT becomes critical: the unit builds, while IT reviews and integrates. Where this works well, IT has an interest in bringing prototypes into operation and helps translate between technical and business perspectives. Where it works less well, IT sees prototypes only once they are finished and can review them but no longer help shape them.
- An AI governance board that brings together the business, IT, HR, the works council and data protection representatives. It sets the framework, resolves fundamental questions once rather than in every project, and provides a forum where employee concerns can be raised early.
Employee co-determination as an organizational issue
Once AI moves beyond making suggestions and starts executing tasks, almost every process raises questions about oversight, data and often jobs. At this threshold, the change management needs also shift. Before scaling, individual employees try AI tools voluntarily. As AI scales, it executes processes across the organization, affecting many employees who may have no choice in the matter. Curiosity turns into concern, and the questions become more concrete: Who oversees what the agent does? What data can it access? What happens to my job?
Employee co-determination at this stage is therefore not just a compliance issue, but an organizational one with three key levers:
- Level: Are negotiations conducted around an overarching framework—for example, a framework agreement on the use of AI agents with clear principles on data, oversight and training—or separately for each use case?
- Timing: Does the works council have a seat on the governance board and see prototypes as soon as they become tangible, or does it receive the finished concept? We explain what this early review of each prototype can look like during a design sprint in AI-first process design.
- Alignment: Do the business function, IT and HR speak with one voice, or do employee representatives encounter three different versions of the initiative? Finding a shared position early helps resolve validation questions much faster and builds momentum for every subsequent project.
HR plays a central role in this framework because it brings together the questions that determine whether scaling works in practice: Which roles are changing? What skills do employees and leaders need? What does leadership look like when part of the team consists of agents? And how is employee participation organized? HR brings these questions into the framework and can bring the works council and data protection representatives to the table early. We explore what this means for the HR function itself in our article on the AI-centric organization.
What differentiates the undconsorten approach?
undconsorten combines the scaling of AI with a clear division of responsibilities among business functions, IT, HR and governance. We help organizations establish cross-functional responsibility for prioritization and process design, build an effective AI unit, and address employee co-determination, data protection and training early in implementation. Rather than remaining an isolated technology project, AI transformation creates an organizational architecture that enables action and innovation, builds trust and supports lasting AI adoption at scale.
Six questions for self-assessment
- Do we have someone with the mandate to set AI priorities across organizational boundaries?
- Who actually builds AI solutions in our organization, and how is the division of work with IT defined?
- What guardrails allow our business areas to experiment independently?
- Do the works council and data protection representatives sit on a body that sets the overall framework, or do they review individual cases?
- Do our leaders understand how their roles will change when agents become part of their teams?
- Will our current division of responsibilities still work when three pilot projects become a program covering ten processes?
Let’s discuss:
How are AI responsibilities divided in your organization, and where are the friction points? We’d be happy to exchange perspectives with you.
