How AI is transforming workflows and redefining roles in companies
Artificial intelligence for answering emails, summarizing documents, or generating code has already become routine at many companies. Most organizations see AI as a one-off productivity tool — something you plug into a specific task and expect it to get done faster. But what if that piecemeal approach is completely overlooking the technology’s greatest potential?
That is exactly what new research from the MIT Sloan School of Management is putting on the table. The study, titled Chaining Tasks, Redefining Work: A Theory of AI Automation, argues that AI’s true impact is not about optimizing isolated tasks, but about how it can reorganize entire workflows — changing the way activities are sequenced, grouped, and distributed between humans and machines.
The research was led by Peyman Shahidi, a doctoral candidate at MIT Sloan, co-authored with Mert Demirer and John Horton, both also from MIT Sloan, along with Nicole Immorlica from Yale University and Microsoft, and Brendan Lucier from Microsoft Research.
In other words, the question companies should be asking is not how can AI help me with this specific task, but rather how can I redesign my workflow to better leverage what AI has to offer.
It might sound like a subtle shift, but it makes all the difference in outcomes. 🚀
What the MIT research actually found
The study from MIT Sloan is not just another surface-level analysis of market trends. The researchers went deep into theory and built models representing how tasks are sequenced and connected in real-world workflows. From there, they reached a conclusion that goes against what most companies practice today.
The dominant logic is still simple: you take a time-consuming task, plug an artificial intelligence tool into it, and the problem is solved. But the data shows that this approach, while it delivers some gains, falls far short of capturing what the technology can truly deliver. The real productivity leap happens when AI is not just wedged into an existing process, but when that process is rethought from scratch — taking into account what AI does well and what still depends on human judgment.
As Shahidi put it pretty directly: the central question is no longer just how AI improves a single task, but understanding its effect at a broader system level, not merely as point productivity tools at the level of each individual task.
The research introduces the concept of task chaining, which is basically the idea that activities within a workflow do not exist independently. They connect, depend on each other, and form a chain. When you automate just one link in that chain without considering the others, the impact is limited. But when you redesign the entire chain — thinking about where AI can take over, where it can assist, and where humans need to stay in control — the result is a real transformation of workflows.
And here is a crucial detail: not all chains are equal. When adjacent tasks are well-suited for AI, they can be grouped efficiently. However, when even a single step is difficult for AI, it can break the entire chain. Shahidi was direct on this point, explaining that if one of the tasks is extremely difficult for AI, that single task will compromise the whole operation.
Another important finding from the study is that this reorganization is not neutral when it comes to impact on people. When workflows change, roles change with them. Some tasks that used to require hours of human effort get done in seconds by AI, while new responsibilities emerge that demand more analytical, creative, and relational capabilities. That is not necessarily a bad thing, but it requires companies to be ready to manage this transition carefully and thoughtfully — from both the technological and human perspective.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The example that makes everything clearer: teachers and tutors
To illustrate how the same skill set can create completely different automation opportunities depending on the workflow, the researchers used a pretty intuitive example: the difference between classroom teaching and one-on-one tutoring.
Both activities involve similar tasks — explaining concepts, creating instructional materials, and assessing student comprehension. But the workflows are quite different. A teacher giving traditional classes prepares content in advance, which allows significant parts of that process to be automated by AI — things like creating slides, structuring lesson plans, and even generating exercises.
A tutor, on the other hand, operates in a continuous back-and-forth flow with the student. Each response depends on the previous one, and every explanation needs to adapt in real time to the learner’s level of understanding. This kind of dynamic and unpredictable interaction severely limits automation opportunities.
As Shahidi explained, the ability to automate workflows using AI is very limited in that second occupation. The way these tasks show up in a profession’s workflow becomes a determining factor.
This example is powerful because it shows that the analysis cannot be done just by looking at tasks in isolation. Two roles with nearly identical activities can have radically different automation potential, simply because of how those activities are organized in day-to-day work.
Why system efficiency matters more than perfecting each step
One of the most counterintuitive findings from the research is that AI does not need to outperform humans at every individual task to generate real value. In practice, organizations can benefit from assigning entire chains of tasks to AI, even when humans could execute some of those steps at a higher quality level.
The reason behind this comes down to coordination costs. Every time work passes from AI to a human and back again, it requires review, validation, and adjustment. These checkpoints slow the system down as a whole. By contrast, letting AI handle a complete end-to-end sequence can eliminate friction, reduce unnecessary handoffs, and speed up final delivery — even if the quality of some individual steps is slightly lower.
Shahidi explained that this gain comes from saving human time, noting that eliminating repetitive oversight can offset marginal performance differences in isolated steps.
This completely changes how leaders should evaluate AI within their organizations. Instead of asking whether AI is better than a human at each step of the process, the question should be: does AI improve the efficiency of the workflow as a whole? This shift in perspective is what separates surface-level adoption from real transformation.
This principle also reinforces the importance of task adjacency. When AI-compatible tasks are grouped together, they can be executed in a single, continuous flow. When they are scattered or interrupted by steps AI cannot handle well, the benefits drop considerably.
The difference between automating tasks and redesigning processes
There is a fundamental distinction that the MIT study makes very clear, and it is more important than it might seem at first glance. Automating a task means using AI to do something faster that a human was already doing. Redesigning a process means questioning whether that task needs to exist the way it does, whether it could be done in a different order, whether it could be combined with another task, or whether it could be eliminated entirely because AI solves the problem in a completely new way.
These are two completely different levels of automation, and the productivity gains between them are equally different. Companies that stay at the first level see some efficiency improvements, but those that reach the second level change the game entirely.
A practical example helps make this clearer. Imagine a marketing team that needs to create campaign performance reports every week. The most common approach today would be using AI to draft the report text based on the data provided — which already saves time. But the process redesign approach would be different. It would question why the report needs to be created manually in the first place, connect data sources directly to an AI-powered system, generate automatic analyses with key insights, and deliver a customized document to each stakeholder with the level of detail they actually need. The end result is not just faster — it is qualitatively superior, because artificial intelligence did not simply replace a step; it transformed the entire process.
This mindset difference also has a direct impact on how teams interact with the technology. When the focus is on automating individual tasks, people tend to see AI as a support tool — something that helps, but does not change the essence of the work. When the focus is on redesigning processes, AI becomes a structural part of how work is organized and executed. This requires a cultural shift that goes beyond training sessions on how to use a new tool — it involves revisiting how the company thinks about productivity, collaboration, and value creation.
How AI is redefining roles and responsibilities
Historically, professional roles have been defined as bundles of tasks that make the most sense when performed by the same person. AI changes that equation by reducing the cost of certain activities and enabling new combinations of work that were not viable before.
For example, if AI can automate several routine tasks within a role, professionals can take on additional responsibilities — often work that requires more judgment, critical analysis, and decision-making. Over time, this reshapes how work is distributed across entire teams and departments.
This dynamic turns AI adoption from a purely technological decision into an organizational design challenge. Choosing the right tool is not enough. Companies need to rethink how teams are structured, which competencies are priorities, and how information flows within the organization.
And this process requires patience. Many companies expect quick returns on their AI investments, but the MIT research indicates that truly meaningful gains only show up after organizations adapt their workflows and build enough capacity to operate in a new way.
As Shahidi pointed out, until that threshold is reached, the costs of AI adoption outweigh the gains. Only beyond that point does the restructuring of work around AI begin to deliver measurable benefits.
Why most companies are still at the basic stage
If the potential of process redesign is so clear, why are most companies still using AI in such a superficial way? The answer involves a combination of factors ranging from pressure for quick results to the real difficulty of mapping and questioning processes that have existed for years.
Many organizations adopt AI tools with the goal of showing they are up to date, but without a clear plan for how to integrate the technology strategically. The result is a collection of point solutions that generate some efficiency but do not come close to what would be possible with a more structured approach to work redesign.
Beyond that, redesigning workflows requires a level of analysis and questioning that goes against the operational logic of many companies. Established processes have inertia. People are used to them, were trained to execute them in a specific way, and often resist changes they perceive as threats to their roles. The MIT study points out that overcoming this resistance is one of the biggest challenges in implementing high-impact automation — and that companies that manage to do this with clear communication and solid planning come out significantly ahead.
There is also a relevant technical component. To redesign processes with AI, you need a reasonably deep understanding of what today’s artificial intelligence systems can actually do, where they are reliable, where they make mistakes, and how they can be combined into more complex sequences. This knowledge is still uncommon among the leadership of many companies, which creates a gap between the technology’s potential and the actual ability to leverage it. Closing that gap is one of the most strategic investments an organization can make right now.
What actually changes when workflows are redesigned
When a company truly adopts the logic of workflow redesign with AI, the results show up across multiple dimensions at the same time. Execution speed increases, but that is not all. The quality of deliverables improves because AI can process volumes of information that would be impossible for a human to handle in a reasonable timeframe. Consistency also increases, since automated processes do not have bad days, do not get distracted, and do not make the same types of human errors.
All of this together creates an environment where people can focus on what truly requires their judgment, creativity, and ability to connect with others, while AI handles the more operational and repetitive layers of the work.
From a productivity standpoint, the numbers can be impressive. The MIT study and other recent research in the field indicate that companies adopting a process redesign approach achieve significantly greater efficiency gains than those simply adding AI tools to existing workflows. In some areas — such as customer service, data analysis, and content development — the difference can be several times over in terms of work volume produced with the same team. This has direct implications for business competitiveness, especially in markets where speed of execution is a key differentiator.
A new principle for work design
The MIT research leads us to a principle that sounds simple when you read it, but has profound implications: how tasks are grouped matters just as much as which tasks are automated.
This means that two companies using the exact same AI technology can get radically different results, simply because of how they organized their processes around that technology. One might see marginal gains, while the other could completely transform its operations.
Work redesign also creates opportunities for teams to develop new skills. With less time spent on repetitive tasks and more room to focus on strategic decisions, people tend to build a deeper understanding of the business, a more systemic view of processes, and a greater ability to spot improvement opportunities.
Organizations that treat AI as a plug-and-play tool may see incremental improvements. Those that rethink how work is structured — grouping AI-compatible tasks, reducing unnecessary handoffs, and redesigning flows — have a much better chance of unlocking the technology’s real potential.
As Shahidi put it very directly: it is not about how to introduce AI into my existing workflow, but about how to redesign my workflow in a way that is more AI-friendly.
That is a virtuous cycle that starts with smart automation and ends with teams that are more capable, more engaged, and more prepared for whatever comes next in the world of artificial intelligence. 💡
