IBA editorial team Is AI merely a tool, or the megatrend that will change everything? To answer this question, we first need to imagine what new forms of collaboration with AI might look like. What is driving the new division of labour between humans and machines in organisations? What does the shift from assistance systems to agentic AI mean for value creation? And which perspectives, roles, work environments and structures will employees need in order to collaborate productively in the future?
On behalf of the German Interior Business Association (IBA), Birgit Gebhardt is currently working on the next New Work Order study, Collaboration with AI, which she will present at ORGATEC in October 2026. At the IBA Forum, she shares initial insights from research, business practice, and trend analysis, combined with pioneering real-world approaches in this pre-read edition of the study.
From AI-integrated to AI-native: When aI becomes part of the value chain
AI-integrated companies such as Microsoft, SAP and Salesforce are currently demonstrating how generative AI and agent tools can be embedded into existing structures to improve performance. However, genuine leaps in productivity only occur when organisations consistently align their structures, roles and processes with collaboration between humans and AI agents. Only then can the potential of this new division of labour be translated into measurable value creation. But how can this be achieved?
To date, corporate and office structures have been designed to organise division of labour, distribute information, monetise knowledge and manage networks. With AI, however, knowledge work is undergoing a fundamental shift: data processing and information exchange are no longer merely organised, but increasingly automated. “Division of labour” now means that machines themselves become part of the value chain – requiring a reassessment of human tasks.
Generative AI and agent systems are emerging at a time of skilled labour shortages and rising performance expectations. They can accelerate processes, reduce workloads and improve outcomes – but only if humans and agents collaborate synergistically. This requires organisations that do not simply layer AI onto existing structures, but instead redesign them according to intelligent value-creation logics.
AI-Assistents vs. AI-Agents
AI-Assistents
- execute tasks upon request (prompt-based)
- act reactively and are usually limited to clearly defined tasks or input–output processes
- provide conversational interfaces and deliver outputs in formats such as text, tables, images, or video
- can be integrated into workflows as automated sub-functions (e.g., customer service)
- are pre-trained using machine learning (foundation models) and fine-tuned for specific tasks
AI-Agents (Agentic AI)
- demonstrate agentic behaviour in goal pursuit, adaptation and contextual awareness
- work towards defined objectives and autonomously develop strategies to achieve them
- act proactively, learn situationally, recognise challenges and adapt to change
- operate iteratively in plan–execute–reflect cycles (e.g., ReAct, AutoGPT, LangChain agents)
- orchestrate complex processes across multiple tools or systems and dynamically integrate APIs and services
Agentic AI: More than the sum of its parts
Agents are not only becoming “digital colleagues”, but also new drivers of value creation. Organisations that own them can manage them internally and monetise them – for example through token systems, by renting out multi-agent systems, or via new business models as agent providers. Organisations without such capabilities risk focusing solely on cost reduction, while actual value creation remains with platform providers – as already seen during earlier waves of digital transformation.
It is therefore no coincidence that the technology and start-up ecosystem is seeing the emergence of the first AI-native companies. These organisations are redesigning their structures entirely for an agentic workforce and expanding their business models as agent providers. Three steps along this path are illustrated in the model: from AI integration to AI architecture, and finally to agent monetisation.
This represents the upper limit of what can be achieved when AI is strategically embedded and consistently aligned with value creation.
How can companies move toward this goal?
Current sentiment is mixed: optimistic expectations of agent systems contrast with growing job cuts. Cost-saving potential is often identified in administrative functions. Lufthansa Group, for example, plans to reduce 4,000 positions. In September 2025 alone, Salesforce cut 4,000 jobs in customer service. Marc Benioff, Chair and CEO of Salesforce, puts it plainly: “We are the last generation of CEOs who manage only humans. We are now moving into a world where humans and agents are managed together.”
This means that, in order to realise synergies between teams and AI agents, work must be redistributed and reorganised. A 2025 MIT survey shows that only five per cent of companies have achieved measurable performance gains so far. While around 80 per cent use AI tools such as chatbots, their impact typically remains limited to individual employees rather than the organisation as a whole.
what needs to happen for humans and machines to collaborate productively?
In companies that have already integrated AI, employee roles are beginning to change. Salesforce, for example, uses AI agents to create synergies and shift employees towards higher-value tasks. Customer service staff move into sales roles with direct customer contact, new business development or after-sales support. COO and CFO Robin Washington announced plans to increase staffing in sales by 22 per cent. At the same time, employees need to develop the ability to assume new roles flexibly across functions in order to collaborate successfully with agents.
This shift requires not only individual adaptation, but also builds on lessons learned from digital transformation. Much of what was introduced there – agile project work, self-directed organisation, cross-functional teams and iterative feedback loops – is already pointing in the right direction. Even if, or perhaps precisely because, agile ways of working often clashed with existing structures, they mobilised those who wanted to achieve goals, simplify processes or try new approaches. As a result, those who actively shape and improve their work benefit – both within their own area of responsibility and across the organisation as a whole
As a result, AI transformation must be driven from both directions: bottom-up, from departments that understand their pain points and can best assess where AI adds value to their work; and top-down, from executive leadership, to establish and clearly communicate a coherent AI strategy that translates into organisational benefits.
A strategic approach to AI means moving beyond isolated pilot projects and embedding AI holistically into organisational structures and processes to enable new business models, more direct routes to market and faster innovation cycles. But how can organisations successfully make this shift?
The study proposes a new value creation model that focuses on the organisation’s core value drivers and aligns structures and processes accordingly. For Germany’s Mittelstand in particular, innovation capability could serve as the primary source of value and be embedded at the heart of the organisation through its research and development function.
networked value creation with innovation at the core
For AI to unlock its full potential, simply integrating tools into existing workflows is not enough. Organisations must rethink value creation – moving away from rigid functional chains towards interconnected structures in which innovation, IT, HR and operational units collaborate more closely. Innovation becomes the core function, linking product development, services and business models. Strategic functions such as strategy and governance, IT and AI infrastructure, finance, legal and HR provide the framework within which AI acts not merely as a support function, but as an enabler that connects processes across the organisation.
For HR, this means that skills development will no longer focus solely on individual job profiles, but on the interaction of roles, skills and agents. HR, IT and leadership must jointly identify which competencies are required, and where, to enable teams to use AI productively.
The model is intended as a framework for discussion on how collaboration with agents can be reorganised. As division of labour and roles change fundamentally, not only agent systems but also organisations and employees require a clear strategic framework and defined objectives. Those who design this collaboration in a value-oriented way can sustainably increase efficiency, innovative capacity and overall performance.
