Skip to main content

The most recent wave of artificial intelligence (AI), spearheaded by the advent and mass adoption of large language models (LLM), showed the potential to fundamentally change how organizations operate and create value. When considering the impact that AI can have on the operating model for teams or companies, three main areas of focus emerge:

  • Workforce: Human-AI augmentation and interfaces that improve efficiency and quality, and open the door to leaner, more streamlined, or alternative staffing arrangements in combination with the two other focus areas in this list.
  • Processes: AI enhances the evolution of Robotic Process Automation (RPA) and Business Process Modeling (BPM) to support faster cycle times, accuracy, and audibility — and reduce bureaucracy.
  • Operations: Leveraging AI-based decision support systems to enrich and enhance ways of working. Concepts like AIOps traditionally applied to IT can be expanded to PeopleOps, Embedded FInance, etc.

One of the most hotly debated aspects of the AI revolution we find ourselves in is the impact it will have on workers and the structure of the workforce of the future. While the idea of automation is nothing new in business, dating at least back to the Industrial Revolution and the advent of assembly lines, AI tools can automate processes beyond the physical realm and inject themselves into decision-making, a task that has thus far been considered inherently human.

In most current organizational structures, the layer that stands to gain the most from reinventing itself in the AI age is middle management. As individual contributors get AI tools that increase their scope and efficiency, and AI-powered process automation takes over administrative tasks, the option to rely on increasingly self-organizing corporate structures becomes more viable.

This will require a fundamental change in how management structures are set up, as workers who are overburdened with administrative tasks — often the middle managers — shift their time and skills to areas of the business where their impact and ability to add value are higher. This is in line with what McKinsey highlights in a recent study on rethinking the role of middle managers.

Workforce

A recent example of this trend can be seen in companies like Bayer AG, the pharmaceutical juggernaut, where nearly half of management and executive positions were cut earlier this year. Their new operating model which they have called “Dynamic Shared Ownership” aims to cut hierarchies, eliminate bureaucracy and accelerate decision-making. While AI and automation are not explicitly mentioned it is easy to imagine that they’ll play a crucial role in enabling this new target operating model. Likewise, “X” the company formerly known as Twitter has gone through a similar restructuring, and while there are varying opinions on the content and usability of X nowadays, purely from a technical and platform infrastructure perspective, X continues to support a growing user base with high concurrency and service levels despite radical headcount reduction.

Output or value generation is crucial in this discussion. Essentially, what is the ROI of your workforce? If we look at this as a simple function of:

Workforce ROI

It is clear that the path toward increasing your workforce ROI is either reducing your workforce or increasing revenue by making your existing workforce more productive. Naturally, a third option emerges: to do both.

These examples speak to a shift in how organizations think about their operational model and in particular the composition of their workforce and their internal operations and processes. If we focus on the workforce aspect, an emerging trend is the workforce composition, and in particular the ratio of management to individual contributors. Something we could call a Management Index (MI) is defined as:

Management Index

There is no preset objective for the MI value here and the sweet spot for each organization may well be different, but the idea is that by leveraging AI, a company could efficiently and effectively sustain a lower ratio, that is, more ICs per Manager.

A clear example of this shift can be seen in Amazon CEO Andy Jassy’s mandate that each organization increases “the ratio of individual contributors to managers by at least 15%” by the end of Q1 2025. According to Jassy, “Having fewer managers will remove layers and flatten organizations…. If we do this work well, it will increase our teammates’ ability to move fast, clarify and invigorate their sense of ownership, drive decision-making closer to the front lines where it most impacts customers (and the business), decrease bureaucracy, and strengthen our organizations’ ability to make customers’ lives better and easier every day.” (Fortune, September 2024)

Jassy’s comments about flattening organizational structures and decreasing bureaucracy are in line with the reasoning behind the moves we mentioned earlier in this blog that companies like Bayer AG, X, and NVIDIA have enacted recently.

Hackman Authority Matrix

While it’s not new, Hackman’s Authority Matrix (Hackman, 1986) provides a good framework for understanding changing workforce dynamics. In the case of Bayer and X, there is a shift from Management toward Teams and more specifically toward self-designing and even self-governing teams. This idea is not new and companies like Spotify and Stripe attempted to move in this direction a decade ago but struggled to do so effectively. So, what has changed?

These different Management-Team structures can coexist within the same organization across business units and in fact, across dimensions of the company e.g., a company as a whole might be seen as management led by a CEO, while specific departments like R&D operate mainly through self-governing teams, and overarching functions like Finance might operate via self-managing teams.

Structures toward the right of the diagram will have a lower Management Index (MI), but reducing this index is not an objective and only becomes valuable when put in the context of value generation. Furthermore, streamlining or restructuring the workforce will not, on its own, have a meaningful impact unless it is accompanied by an equivalent change in process and operations that enhances the new company structures.

Here is where AI can play a significant role and be the catalyst: enabling automation and augmentation in day-to-day operations, reducing time to insight, providing decision support systems that help teams make better and timely decisions, and facilitating processes that reduce bureaucracy.

Processes

There are many examples of processes being defined based on their complexity. One such definition is that a complex process is characterized by its intricacy, involving numerous decision points, conditional pathways, and interactions across various departments or stakeholders.

When we examine organizational processes in terms of complexity, we can simplify this complexity by considering the number of tasks involved and assigning a continuous complexity value on a scale from 1 to 5 for each task. To illustrate this further, think of the complexity value for a task as analogous to the Richter scale for measuring seismic magnitude, which is logarithmic. This means that a level 5 task is ten times more complex than a level 4 task.

The next step is to estimate how many of these tasks could be automated, omitted, or simplified through the use of AI, either by reducing complexity or enhancing execution.

For a given process its Process Complexity Index (PCI) could be calculated as

Process Complexity Index

The role of AI here is to reduce the number of tasks and/or reduce the complexity of individual tasks in the chain.

Organizational processes, particularly those related to administrative tasks such as filling out forms, managing sheets, and handling approvals and reviews, are well-suited for incorporating AI. The combination of enhanced process automation aided by AI with flatter and leaner organizational structures has the potential to dramatically reduce red tape and bureaucracy.

An important aspect that emerges is the need to properly orchestrate and link these processes, an element that today is often hindered by disconnected flows, manual intervention, and lack of end-to-end traceability. Process mining, which has enjoyed some popularity in recent years, is a critical capability to identify bottlenecks, and effectively materialize and write down processes that are implicit and tacit, deeply rooted in organizational customs and culture. Making processes explicit and being able to point to specific links in the chain will make it more likely that we can design AI agents that are able to automate and augment (parts of) the process at hand.

If we look at the most recent discussions around GenAI, we see an increasing focus on how to chain, orchestrate, and combine multiple smaller, specialized agents each with a dedicated task. This sits in stark contrast to the alternative of having ever larger monolithic models and signals a shift toward separating existing processes into tasks that specialized, dedicated AI agents can collaborate on and complete together. Imagine a scenario where an Agentic AI system follows a certain process and self-identifies steps that are candidates for automation, and (automatically) creates an AI agent to perform it. Today, companies like Orby are already developing this kind of functionality.

Interestingly, a side effect of AI-driven process design may be that individual contributors and those responsible for re-engineering and implementing these new AI-driven processes will need to develop skills traditionally associated with management (Harvard Business Review, 2024). This includes abilities such as delegation, providing a clear scope of work, and allocating budgets and objectives to effectively guide and instruct the AI systems involved in the latest wave of Generative AI Process Automation, also known as Augmentation.

Operations

When discussing operations, we are referring to the way things are done and executed, which often requires the combination of people and processes. There are various levels and dimensions to consider when discussing the impact of AI on operations. For instance, as mentioned in the workforce section, different departments within a company may be better suited for specific team configurations. Additionally, the way individual departments function can vary based on their composition, objectives, and scope.

The role of AI in operations is ultimately to facilitate decision-making by streamlining the interaction between people and processes. An area that has gained significant traction over recent years is AIOps, essentially the use of AI (particularly in IT) to optimize infrastructure and application management.

IT was perhaps an obvious first choice due to its often deterministic nature and the possibility of encoding a lot of the required decision-making into neat mathematical constructs. For example, training a model that helps predict and analyze the impact of increased I/O on your server infrastructure can be quantitatively parametrized.

However, the advent of GenAI with large language models, multimodal frameworks for image-to-text, text-to-speech, and vice versa opened the door to embedding and leveraging AI as a decision support system for more qualitative operations that had been purely or primarily in the human domain. Areas like PeopleOps, HROps, (Embedded) Finance, and many others should now be considered candidates for the next wave of AIOps-like frameworks.

Operations also extend to factors beyond the internal scope of organizations. Critical operative decisions such as out or near-sourcing, offshoring, and partnerships will be impacted by the use of AI. Supply chain operations is one area where processes can benefit from automation and augmentation. This is especially true in scenarios that require extensive coordination and communication between different systems. The use of compound AI systems can significantly enhance these processes.

Conclusions

In this blog, we’ve focused on three key organizational topics: Workforce, Process and Operations, and the impact that GenAI brings to each. Companies of every size and shape should think about how to best integrate and leverage these new technologies to augment and elevate themselves. As with most things, there is no one-size-fits-all or silver bullet as to where and how a company should use AI or how to implement it - but it’s crucial to focus on:

  • Re-thinking organizational structures: How can AI augment your workforce and extend focus to high-value objectives?
  • AI-driven process augmentation: What processes in your organization can be automated, simplified, or streamlined by using AI agents, shifting focus from individual use cases to end-to-end processes?
  • Operate better with less: How can you reduce levers or processes to use resources more effectively and efficiently, make timely decisions, and take effective actions?

To learn more about how Databricks can support your AI and Data Strategy with the Databricks Data Intelligence Platform take a look here.

Try Databricks for free

Related posts

How Databricks enables your operating model for Data and AI: Part 1

April 26, 2023 by Fabian Lanz in
"The bottleneck (for AI) now is in management, implementation, and business imagination, not technology." -Erik Brynjolfsson Data and AI have emerged as strategic...

How to execute your operating model for Data and AI

August 8, 2023 by Fabian Lanz in
In Part 1 of this blog series, we discussed how Databricks enables organizations to develop, manage and operate processes that extract value from...

Data + AI Strategy: People Focus

This post is part of a series. Check out Part 1: The Data + AI Trifecta: People, Process, and Platform In the current...

The Data + AI Trifecta: Process, People, and Platform

There's rapid demand building among enterprises to build and leverage Generative AI applications. In fact, according to a recent McKinsey study, one-third of...
See all Data Strategy posts