Much of the debate about artificial intelligence and the future of work asks a familiar question: Which human tasks will machines automate? It’s an important question, but it captures only part of what is happening.  

Technology not only changes the work organisations perform, but also how they organise that work – even before AI entered the scene. A manufacturing robot is primarily a production technology that changes how a product is made. But a technology that allocates tasks, routes information, monitors progress and helps coordinate workers changes how the organisation itself works. We refer to the latter as “OrgTech” to distinguish it from production technologies.  

AI is increasingly performing both functions. Consider an AI agent that helps a salesperson prepare a client proposal. Used this way, AI primarily changes production: The salesperson can produce the proposal faster or better. Now imagine an agent that notices the proposal requires input from legal and finance departments, identifies the relevant colleagues, routes the required information to them, reconciles their responses and alerts the salesperson if something needs to be resolved. The technology hasn’t simply automated another production task; it has absorbed part of the organisation's coordination architecture.  

This distinction matters because the organisational consequences can be very different. 

OrgTech can be vertical or lateral  

At its core, every organisation, regardless of what it produces, must solve a recurring set of problems. Goals must be broken down into tasks and allocated to people who need both information and motivation to perform those tasks. And somebody – or something – must deal with exceptional cases that inevitably arise. 

There are different ways to solve these problems. Organisations can rely on managerial authority acting top down (vertically), whereby managers allocate resources, monitor performance, coordinate activities and intervene when collaboration breaks down. Or they can rely on peer-to-peer (lateral) coordination among their subordinates, whereby employees solve problems through information sharing, shared norms and mutual adjustment. 

Enter OrgTech, and we see how these vertical and lateral mechanisms are strengthened by technology. When applied vertically, in the form of better information, monitoring and decision-support systems, OrgTech can increase managerial capacity by allowing managers to coordinate more people or handle more decisions. And if a manager who once supervised six employees can now effectively manage 12, the organisation may need fewer managers for the same staff strength. 

Lateral OrgTech, on the other hand, increases collaborative capacity by allowing employees to coordinate directly with one another with minimal managerial intervention. AI agents that route information, track dependencies, prevent miscommunication and ensure things don’t fall through the cracks could extend these collaborative capabilities considerably. Even before the advent of AI, collaboration tools such as Jira, Asana and Slack already existed.  

Our research on the adoption of such collaborative work-management technologies across roughly 3,000 large United States-based public firms from 2010 to 2019 shows that following adoption, firms experienced a decline in the number of managers per employee (managerial intensity). At the same time, job advertisements placed greater emphasis on delegation and lateral coordination skills among non-managers. The pattern was clear: OrgTech was shifting some coordination work away from managers towards peer-to-peer interaction – even before its latest AI-based incarnations. 

How OrgTech shapes organisations 

Two organisations can look equally flat on an organisation chart while operating very differently. Here’s how: 

OrgTech can flatten, but may not decentralise 

By strengthening peer-to-peer coordination, lateral OrgTech can lead to both the flattening and decentralisation of an organisation. Managers can intervene less, simply because they don’t need to as often. That is what we found in our research on the adoption of collaboration tools like Jira, Asana and Slack.  

In contrast, vertical OrgTech can lead to a flatter organisation, but not necessarily a more decentralised one. AI-enabled monitoring and decision systems can allow a smaller number of managers to exercise greater control over a larger number of employees, such that the hierarchy may become flatter while authority remains concentrated or becomes even more concentrated. 

OrgTech may increase or decrease the demand for managers  

Then there’s the impact on the need for managerial work. By raising the productivity of managers, vertical OrgTech should reduce the number of managers needed if the scale and scope of activity remain constant. But greater managerial productivity can also make managerial intervention cheaper and more attractive. As such, organisations may monitor more, centralise more decisions, grow larger or take on more complex work. If so, the resulting demand for managerial problem-solving could grow to the extent that more managers are needed.  

Lateral OrgTech works differently. Holding scale and scope constant, it directly reduces the demand for managerial intervention because employees solve more coordination problems themselves. But as in the case of vertical OrgTech, there can be unexpected effects on managerial jobs. If cheaper peer coordination allows the organisation to become much larger or more complex, the absolute number of managers could still rise even while the number of managers per employee (i.e. managerial intensity) falls. 

The more important distinction is therefore about managerial intensity, not managerial headcount. Lateral OrgTech substitutes directly for managerial intervention and should therefore reduce managerial intensity. Vertical OrgTech makes managerial intervention more productive – and potentially more attractive – so its effect on managerial intensity is ambiguous. 

Technology doesn’t determine the organisation 

None of this means that there is something inevitable about what AI will do to organisations, managerial headcount or even managerial intensity. Remember, the same AI technology (predictive, generative, agentic, etc.) can be deployed as a production technology, vertical OrgTech or lateral OrgTech. This deployment decision changes everything.  

That’s why we think only one possible future resembles a digitally enhanced version of Tayloristic “scientific management”: fewer layers, much larger spans of control, extensive measurement and monitoring, coupled with highly concentrated authority. Another future, when using the same technological capabilities to make peer-to-peer collaboration easier, could look like this: employees have greater access to information, coordinate directly, resolve more exceptions themselves and depend less on managerial authority. Both futures involve flatter organisations, but only the second produces greater decentralisation and lower managerial intensity. 

Ultimately, which future emerges won’t be determined by AI capabilities alone. It will depend on which production and organisational problems leaders choose to use those capabilities to solve, and what policymakers incentivise and disincentivise. Once we recognise that, the central managerial question moves beyond “What jobs can AI replace?” to a more consequential one: What kinds of organisations can AI enable us to build?