TL;DR. AI is dissolving settled categories: businesses, jobs, deliverables, prices, and timelines that we have organised around for thirty years. Two phases run in parallel. Schumpeter's destruction, accelerated by agentic engineering and faster than any prior cycle. And a quieter rebuild that economists call institutional entrepreneurship: someone has to define what the new roles, deliverables, and standards will be. The job for Mittelstand boards and family offices is to spot which categories in their business are dissolving and shape what replaces them. History suggests the window is two to five years.
The Settled World Is Unsettling
Last week I designed five custom AI agents for a Swiss client workshop, threw the architecture away the night before, and replaced it with standard Claude Code deployments delivering the same outcomes for a fraction of the build. The custom agents had been a comfortable detour from the real task: redesigning what the firm actually does in 2026.
That detour has a name.
Two Phases Are Running at the Same Time
Schumpeter called the first phase creative destruction: the displacement of old structures by new ones. Fast, visible, and the part everyone is talking about. The second phase has its own name: institutional entrepreneurship. Slow, contested, and led by specific people who define what the new categories mean, who counts as legitimate, and what good work looks like.
Before Google, there was no SEO. Today there are degree programmes, vice presidents of online marketing, and heads of affiliate. The infrastructure looks inevitable in retrospect. It was not. Specific people, broadly between 1998 and 2005, made it inevitable by treating an undefined area as their professional opportunity.
The same window is open now in agentic engineering. There is no settled definition of what an AI-augmented consulting practice contains, what an AI Operations function reports against, who certifies a vibe coding (AI-assisted rapid prototyping) practitioner, or how a CFO should account for the productivity dividend. History suggests these questions get settled within five years. None of them have an answer today.
The discomfort of working without a category to claim is the work itself.
What This Means for Your Next Quarter
The Mittelstand and family-office advantage has always been the willingness to rebuild a business slowly, in public, without consulting decks or McKinsey backup. That willingness is rare and unevenly distributed. It is also exactly what institutional entrepreneurship requires.
Three concrete moves.
First, identify one category in your business in active dissolution. Procurement may no longer mean RFPs and SOWs. Customer support may no longer mean tickets and SLAs. R&D may no longer mean stage-gate reviews. The early signal is that existing role descriptions feel slightly absurd when read against what the team is actually doing.
Second, appoint someone to redefine that category. Not to optimise the existing function. To rewrite its job description, deliverables, metrics, and how it connects with the rest of the business. Treat the role as quarterly iterative for the first year, with explicit permission to throw away and restart. Build the team around claude code workflows and vibe coding practices, not around the existing skill matrix.
Third, accept that the answer will look uncomfortable to anyone benchmarking against three-year-old norms. That discomfort is the signal. If the new shape looks like the old shape with agentic engineering bolted on top, the redefinition has not happened yet.
Where This Argument Bends
Not every dissolving category will rebuild. Some genuinely die. The first wave of "Web Masters" in the late 1990s mostly did not become anything stable. Most early-2000s "social media gurus" also did not. Survivor bias makes institutional entrepreneurship look more rewarding in retrospect than it is in the moment.
Timing matters more than category-spotting. Being first to a category that takes ten years to settle is functionally equivalent to being wrong, if the business cannot survive ten years of waiting. Mittelstand and family-office capital structures help here, because patient capital tolerates institutional uncertainty better than venture or quarterly-public structures do.
The deeper risk is not that the rebuild fails. It is that the new categories settle around different actors than the ones currently positioning to claim them.
What I'm Reading
Paul DiMaggio, Interest and Agency in Institutional Theory (1988). The paper that gave institutional entrepreneurship its name. The core argument: institutions look natural in retrospect because the entrepreneurs who built them succeeded, and the unsuccessful attempts were forgotten. Useful as a framework even if you skip the academic apparatus.
Andy Grove, Only the Paranoid Survive (1996). Grove's strategic-inflection-point framework predates the academic literature on institutional entrepreneurship by a decade and reads more practically. The line that matters: a 10x change makes most of what you know obsolete in eighteen months. Written by an operator, not an academic, which shows.
The destruction phase is loud. The construction phase is quiet, longer, and where the returns sit. The next time you find yourself defending an existing category description, ask whether you are protecting it or whether it is protecting you.
Want to identify which categories in your business are dissolving and where the rebuild is open? Visit drfloriansteiner.com/angebote.