Michael Goshka is founder and CEO of Planfix, an AI-powered teamwork management platform giving every team its own workspace in one system.
getty
Not long ago, realizing that you were talking to a bot instead of a customer support agent was incredibly frustrating. Conversations were highly predictable: Choose an option, repeat your question, fail to get an answer and eventually ask for a human.
But that experience is changing quickly. AI can now be used to draft emails, summarize meetings, analyze documents, classify requests and help people make everyday decisions. We increasingly use it at work and in our personal lives without much thought about AI’s involvement.
While building a SaaS company, I’ve watched the same shift happen inside businesses. However, as AI becomes better at understanding context and handling routine decisions, it raises an uncomfortable question: If technology can coordinate more of the work, do companies still need the managers who did so in the past?
I believe the answer is yes. But I also believe that the value of these managers is changing. AI isn’t eliminating the need for management; it’s just reducing many of the manual tasks we used to confuse with management.
The human router is disappearing.
For years, workflow automation was largely deterministic: When a request arrives, create a task; when its status changes, notify someone; when a deadline approaches, send a reminder. But while that model removed repetitive actions, humans still had to do much of the thinking between those steps. Someone had to read the request, understand what the customer wanted, classify it, determine its urgency and decide who should handle it.
The abilities of modern automation change that model. Consider a typical customer request: Previously, automation might create a task, but an employee still had to interpret the message and determine what should happen next. A manager then monitored progress and intervened if the request stalled.
Today, AI can analyze an unstructured message, extract relevant information, classify its intent and urgency and prepare a response. That information can then trigger a workflow that assigns the appropriate specialist, sets a deadline, monitors progress and escalates the issue when an exception occurs. The result isn’t simply a faster process—several manual coordination steps disappear altogether. This can give employees and managers more time for work that requires judgment, relationship-building and handling exceptions.
But if AI can understand the request and workflows can route the work, the manager’s role shifts from coordinating individual tasks to designing and improving the system that keeps work moving.
Good managers already carry the workflow in their heads.
In my experience, a strong manager is one who knows things such as which customer requires additional attention, when the standard procedure is inappropriate, which specialist should handle an unusual situation and when a minor issue needs to be escalated. They understand the exceptions that no process diagram can fully capture.
In other words, much of the workflow already exists. It simply lives inside the manager’s head.
The problem is that this can create an operational risk. If a process works perfectly only as long as one particular manager is available, what happens when that person takes time off, changes roles or becomes overloaded? Decisions slow down and exceptions accumulate.
That isn’t an argument for replacing the manager, but for making their expertise scalable. The opportunity I see in AI is to separate routine managerial decisions from those that genuinely require experience and judgment. A manager shouldn’t have to make the same routine or prioritization decision hundreds of times if the reasoning is already understood. Their value is greater when they define how those decisions should be made and improve the system when the rules no longer work.
Build around your best managers.
Rather than attempt to automate strong managers, the goal should be to use AI and workflows to increase their leverage. In my experience, making this work in practice comes down to five key principles:
1. Capture decision logic. Identify how experienced managers prioritize requests, assign work, recognize risk and determine when escalation is necessary. If the reasoning is consistent, you can often encode that logic into a rule or AI-assisted workflow.
2. Automate the normal path. Routine situations shouldn’t require constant managerial intervention. Let automation handle predictable routing, deadlines and notifications while AI handles interpretation and classification.
3. Define the human boundary. It is important to decide where automation ends. Some actions may run independently, but others should require approval. Make sure the decisions involving unusual risk, sensitive customers or significant trade-offs remain human responsibilities.
4. Learn from exceptions. When managers repeatedly intervene in the same situation, treat it as process feedback. A recurring exception is often a process waiting to be redesigned.
5. Measure managerial leverage differently. Instead of focusing on how many tasks a manager supervises, consider how much high-quality work the team can complete without their direct involvement.
I’ve found that the last point is particularly important. The best manager isn’t necessarily involved in the most decisions. Often, they’re the one whose team can consistently make good decisions without them.
Aim for better managers and fewer bottlenecks.
When routine coordination no longer consumes most of their attention, one strong manager can support a much larger operation. In my experiences, this means that processes can become less dependent on specific individuals, decisions can become more consistent and problems can be escalated according to clear rules rather than depending on whoever happens to notice them first.
Most importantly, you can give your managers more time for the work automation still struggles with, such as navigating ambiguity, coaching people, handling unusual situations, improving processes and deciding when the existing rules should no longer apply.
Middle management isn’t disappearing. The definition of good management is simply changing. I believe the managers who remain most valuable in the coming years won’t be the ones who route the most tasks or request the most status updates, but the ones whose experiences can fuel systems that help the organization make better decisions.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?
