Four roles that make AI transformation actionable
Why do so many AI initiatives stall despite available technology and successful pilot projects? The new open-access article “From Promoters to AI Role Units” shows that effective AI transformation requires more than suitable applications. Organisations also need clearly defined functions, responsibilities and competencies. The resulting role and competency model provides an important research foundation for FITLO’s services. (= Four roles that make AI transformation actionable)
When technology is not the real problem. Four roles that make AI transformation actionable
Many organisations are experimenting with artificial intelligence, developing initial use cases and implementing prototypes. Yet they often struggle to embed these initiatives in processes, responsibilities and decision-making structures.
The barriers are not purely technological. In practice, organisations frequently face other challenges:
AI initiatives lack a clear mandate or sufficient strategic priority.
Business requirements are not translated into technically testable use cases.
Data access, approvals, governance and operating models remain unresolved.
Business units, IT, leadership and other stakeholders struggle to establish a shared understanding.
Roles and responsibilities are not clearly assigned.
Competency profiles remain too general and are not linked to specific functions.
The article therefore addresses an organisational question: Which functions must be covered during AI transformation, and which competencies are required to perform them?
How the model was developed
The study combines a qualitative case study with an external competency assessment.
Several AI initiatives within a manufacturing organisation were examined over a six-month period. The empirical material included 18 strategy sessions, workshops, coordination meetings and reflection formats, as well as 14 project-related documents. The analysis focused on recurring implementation barriers, contributions made by different stakeholder groups and the functions these contributions performed in moving AI initiatives forward.
Four recurring role units were initially reconstructed from the empirical observations. Only afterwards were they theoretically related to the promoter model from innovation research.
In a second step, 17 external experts from academia and business assessed the relevance of twelve AI competency fields for the four role units. The two-round Delphi process made it possible to prioritise a distinct competency profile for each function.
The four AI role units
Orchestrator
The Orchestrator creates strategic legitimacy for an AI initiative. This role clarifies objectives, priorities, resources, responsibilities and success criteria. Its central function is to establish the mandate and decision-making scope that enable others to act.
When this function is missing, AI initiatives often remain non-binding, lack resources or become caught between conflicting expectations regarding value, risk and implementation speed.
The most relevant competencies include decision-making, critical thinking, ethical competence, communication and self-determination.
Deliverer
The Deliverer translates business challenges into technically actionable AI use cases. This role specifies requirements, assesses data availability and develops prototypes or solution concepts.
Without this function, use cases remain abstract or cannot be evaluated and implemented on a sound technical basis.
The competency profile focuses on system design, implementation capability, creative problem-solving, critical digital competence and proactive steering.
Enabler
The Enabler creates the organisational and technical conditions required for responsible implementation and scaling. This includes data access, IT architecture, data protection, information security, approval processes, operating models and capability development.
Without this function, even technically convincing solutions may fail because of organisational obstacles.
The Enabler particularly requires proactive steering, system design competence, critical digital competence, collaboration and ethical competence.
Connector
The Connector links different perspectives and mediates between business units, IT, leadership, governance functions and other affected stakeholders. This role translates requirements, addresses concerns and makes conflicts of interest or understanding manageable.
When this function is missing, siloed thinking, resistance and conflicting expectations concerning the possibilities and limitations of AI can obstruct implementation.
The most important competency fields include communication, collaboration, ethical competence, self-determination and critical thinking.
The key insight: these are not four new job titles
Orchestrator, Deliverer, Enabler and Connector are not fixed job titles, nor does the model suggest that organisations must create four additional positions.
They describe clusters of functions that must be effectively covered within an AI initiative. In smaller projects, one person may perform several functions. In larger or more highly regulated organisations, a single function may be distributed across several individuals or existing organisational units.
The decisive question is therefore not whether these titles appear in the organisation chart. What matters is whether strategic legitimacy, business and technical specification, organisational enablement and social mediation are actually present.
Participation requires structure
A further finding concerns the involvement of different stakeholder groups. Participatory AI transformation does not arise simply because as many people as possible attend as many meetings as possible.
Effective participation means involving the right stakeholders in decision-making, implementation, enablement and mediation according to their expertise, responsibilities and degree of affectedness.
The model therefore connects role clarity, competency development, governance and participation. It demonstrates that AI transformation is a distributed task involving both design and responsibility.
How organisations can use the model
The role and competency model can serve as a diagnostic and design instrument. Organisations can use it to examine:
Which of the four functions are already covered in our AI initiative?
Where do gaps in responsibility or competency exist?
Which stakeholder groups need to be involved?
Which competencies should be developed systematically?
Do we need internal role clarification, capability development or an external appointment?
How should responsibilities be distributed across piloting, scaling and ongoing operations?
The model therefore supports not only the analysis of existing AI initiatives, but also project staffing, capability development and organisational embedding.
A research foundation for FITLO’s work
The model provides a particularly important foundation for FITLO Search. Our search process does not begin with a conventional job title. It begins by identifying which function is missing from a specific AI initiative. This forms the basis for defining responsibilities, competency requirements and suitable search profiles.
The role architecture is also relevant to FITLO Research and FITLO Funding. Research and funding projects require more than a compelling research question. They also need a viable project organisation. The model helps clarify responsibilities, partner contributions and required competencies at an early stage.
It therefore illustrates what FITLO stands for: translating research findings into practical instruments that organisations can use to make decisions, structure initiatives and fill critical roles.
Limitations and further research
The role model was developed from an embedded case study in a manufacturing organisation. It therefore does not support statistical generalisation. Instead, the four role units should be understood as an empirically grounded and theoretically consolidated design logic.
Further research should examine how the role architecture applies across different industries, types of organisations and levels of AI maturity. Future studies should also investigate how role and competency requirements change between exploration, piloting, scaling and routine operations.
Read the open-access article
The full article has been published in HMD – Praxis der Wirtschaftsinformatik:
Florian Härer: “Von Promotoren zu KI-Rolleneinheiten: Ein rollen- und kompetenzbasiertes Modell für partizipative KI-Transformation”
AI transformation does not succeed because of one individual role. It succeeds when complementary functions, clear responsibilities and the right competencies work together effectively.