Last Updated on 13. April 2026
All companies are currently working on AI pilots, have defined use cases, are deploying tools, and some are already incorporating AI into their workflows or working with AI agents. The expectation is clear: to become faster, work more efficiently, and solve current problems. But in reality, when AI actually works, a pattern emerges that many did not anticipate. It is no longer just about efficiency. Our thesis: AI does not automatically accelerate work, but rather ruthlessly exposes where the organization is unclear.
Why AI is different from previous technology
Many are familiar with this saying from digital transformation: If you digitize a bad process, you end up with a bad digital process. That’s true. But with AI, a new dimension has emerged.
Previous technology executed what was specified. It automated defined processes and was, in the best sense, a tool—predictable, controllable, passive. We were the thinkers; technology was the tool. If the process definition was correct, the automation worked. If not, it failed at exactly that point, but it failed in a predictable way.
AI does not merely execute. It generates suggestions, recognizes patterns, and interprets context; therefore, it makes its own assumptions in ambiguous situations. When goals, criteria, or responsibilities are not clearly defined, AI interprets these gaps and, in cases of doubt, thereby amplifies the existing ambiguity.
AI acts first as a mirror
Before AI boosts productivity, it first acts as a mirror, revealing unclear decision-making paths, implicit knowledge, and missing responsibilities. This moment of visibility is unexpected for many organizations and represents a major opportunity.
When AI automatically prepares decision templates and makes options for action transparent, it eliminates a buffer that many organizations underestimate. AI eliminates this buffer. What remains are the actual conflicts: unclear responsibilities, conflicting goals, missing priorities. This is uncomfortable, but valuable. Because in these cases, AI hasn’t optimized anything; it has simply made reality visible.
When more output doesn’t mean faster
How this pattern manifests in practice can be illustrated with a clear example from marketing. In the past, the bottleneck was often: Do we have enough ideas? Do we have enough capacity to even create drafts? Output came so slowly that you could muddle through. Coordination and approvals ran in the background because there were only a few options on the table anyway.
Then came AI. Suddenly, ten visuals are created in an hour, campaign drafts in two to three hours. And now something interesting is happening: The bottleneck is no longer creation. Nor is it quality. The bottleneck thus shifts from execution to decision-making. Which variant will be pursued, and based on what criteria? AI has delivered, and human coordination is now the bottleneck.
When transparency forces a decision
The same pattern emerges in a completely different context—and there it becomes strategically relevant. Let’s take a coordination meeting on the project portfolio. A prestige project is on the agenda. Technically, the status should be set to yellow, but it remains green. In the old world, participants would have recognized the broad room for interpretation and moved on.
With AI, project statuses can be evaluated automatically: timelines, budgets, risk indicators. The project is listed on an equal footing with others. Suddenly, it becomes clear that the timeline is yellow, the budget is green, but the risk assessment is red. AI hasn’t optimized anything. It reduces the room for interpretation and forces the organization to talk about reality differently and make explicit decisions. And this pressure falls on the executives.
Decisions aren’t made easier by AI; they’re needed faster.
AI shifts from a mirror to an amplifier
When organizations create this clarity—regarding decision-making scope, responsibilities, and prioritization criteria—then AI can shift from a mirror to an amplifier and thus actually generate productivity.
People remain crucial as the ones setting the direction, whose judgment is indispensable precisely where conflicting goals must be weighed and responsibility must be assumed. The true benefit of AI only emerges once it is clearly defined what is to be achieved and according to which criteria decisions are made—because AI only has an impact when it specifically supports this human judgment. If this organizational clarity is not developed alongside the technology, one is merely digitizing existing problems.
Where companies should start
The most productive question a company can ask itself in the context of AI is not: Where are the greatest efficiency potentials?
Three questions are crucial:
- Where are decisions stalling today?
- Where do conflicting goals remain unresolved?
- What needs to be clarified so that AI can have an impact at all?
mgm consulting partners supports companies in establishing precisely this steering capability as a practical prerequisite for the effective use of AI.
Together with our clients, we analyze where AI can actually have an impact and identify these decision-making bottlenecks or create the organizational conditions so that AI not only functions technically but actually has an impact.
Why this ability to steer is becoming a central organizational competency, especially in dynamic times—and what misconceptions companies must overcome in the process—is described by my colleague Benedikt Jost in his article “Steering in Dynamic Times: How Orientation Translates into Effectiveness.”
This article is based on a presentation by Antje von Garrel, Manager and Lead AI Consulting at mgm consulting partners, as part of the joint keynote and networking event “Effectively Connecting Spaces, Organization, and AI” with Steelcase on March 18, 2026, in Hamburg.
