AI is not just improving software — it is exposing its obsolescence. A deep dive into the "viscosity tax" that paralyzes your company's productivity.
Most legacy software vendors believe they have solved the AI equation by grafting "Copilots" and prompt bars onto their existing interfaces. This is a fundamental misdiagnosis. As McKinsey's recent report on the "AI-centric imperative" makes clear, this is not a routine feature-addition cycle but a major architectural break.
The numbers are unforgiving. According to research firm Deepstar Strategic, "AI-native" startups are now posting a median growth rate of 100%, against a modest 23% for traditional software vendors. More striking still, they reach the $100M ARR mark in one or two years with fewer than 20 employees — pulverizing the standards of the previous decade.
This spectacular velocity is the symptom of a complete overhaul of the value proposition: we are moving from workflow optimization tools to autonomous decision and execution engines. Beyond the technology war, a genuine organizational earthquake is brewing for customer companies.
For years, "technical debt" was the unchallenged scapegoat of IT departments — the systematic justification for slow time-to-market and sluggish innovation. Today, the diagnosis has shifted. The code in these enormous legacy platforms is not necessarily defective; it is their fundamental architecture that freezes the company in place.
These software giants were designed for a world where the human is the exclusive pilot and the tool a passive facilitator. The real threat hanging over customer companies is no longer the technical debt of their vendors — it is the organizational debt those vendors reveal and impose. A genuine paradigm debt.
Trying to turn these click-driven cathedrals into agent-steered systems is like trying to turn a freight train into an airliner. This is exactly the warning Bain & Company has issued: agentic AI threatens to disrupt legacy models by autonomously replicating entire workflows. Traditional players are caught in a vise: fierce competition from ultra-lean new entrants, and end customers starting to internalize their own solutions.
To grasp the violence of this shift, you have to watch how this obsolescence strikes every department of the enterprise simultaneously.
Design and UX were the first laboratories of this rupture. Until recently, tools like Figma reigned supreme. Today, AI-native solutions bypass the "canvas" paradigm entirely. That is the case with Paper (paper.design), which after a closely watched seed round positions itself head-on against traditional workflows, or with Uizard. The user no longer manipulates vectors; they dictate a textual intent or submit a sketch that the tool turns into a functional prototype. In the most advanced workflows of 2026, the UI design software stage even disappears in favor of IDEs like Cursor coupled with AI agents — going from idea to code in real time.
But this dynamic spreads far beyond the creative sphere. In sales, the traditional CRM is fundamentally just an inert database demanding hours of data entry. Against that rigidity, autonomous AI agents prospect, qualify and book meetings without any human interface in between. In customer support, ticket factories are short-circuited by agents capable of resolving incidents end-to-end.
As Saastr recently pointed out, the gap between classic B2B vendors and these new "AI-native" players has become a painful chasm. Software is no longer a passive container: it has become the actor of execution.
The peril does not weigh on the financial health of legacy vendors alone — it weighs on the survival of their customers. There is a tacit law of organizational design:
the tool shapes the process, which in turn shapes the structure of the company.
Historically, deploying a major enterprise software package required the creation of rigid processes and approval silos. What was once an acceptable governance structure now reveals itself as a leaden corset. This is the major negative externality of legacy software: it acts as an organizational toxin. By imposing its architectural limits on its customers, it atrophies their capacity to adapt.
Companies anchored in those ecosystems will pay a genuine "viscosity tax." Their processes, modeled on the limits of obsolete tools, will prevent them from matching the velocity of competitors born with AI — competitors capable of iterating at a speed mathematically impossible to reproduce on a traditional software stack.
This technological mutation produces an unexpected and massive organizational phenomenon: the silent rebellion of the end user. Suffocated by the viscosity tax of IT-blessed tools, the operational employee emancipates themselves.
Access to AI in natural language (the famous "vibe coding," or development by intent) now lets any employee build their own custom tools. An employee facing a repetitive task no longer waits for the company to roll out an official module in eighteen months. They orchestrate their own workflow across several AI agents to complete the mission in minutes.
We are moving from an era of top-down optimization, where the process was carved into the marble of major enterprise software, to an era of bottom-up optimization. Thousands of micro-workflows emerge from the base — ultra-specialized, agile, and disposable once the task is done. Companies that insist on forcing their talent through the tunnels of old software will not just lose productivity; they will see their best people leave — or worse, develop an AI-fueled "Shadow IT" entirely beyond their control.
Faced with this threat, the reflex response of Tech giants has always been acquisition. If you can't build it, buy it. Yet generative AI has broken this defensive mechanism for two major reasons:
We must stop treating AI as just another software feature. AI represents an unprecedented shock to the means of production themselves, because it triggers a vertiginous fall in the marginal cost of expertise. A junior profile, equipped with a well-orchestrated AI agent, can now produce the value of an entire department from three years ago.
The consequences are binary. Organizations that refuse to update themselves at depth — by radically rethinking their processes, their roles, and their value-creation circuits — will not simply fall behind technologically. By keeping cascading validation chains and strata of coordination just to move information around, they will mathematically end up crashing under their own weight.
AI is not an option to add to the company's software menu. It is the new operating system of value creation, and it tolerates no viscosity.
Analysis
The era of the interface as a "cover-up" for organizational complexity is ending. With the rise of the agentic, design does not disappear: it shifts from the surface of the screen to the structure of the contract. The click no longer carries authority — semantic clarity does. A case for moving from designing journeys to architecting robust business models.
Analysis
Organizations invested millions to deploy AI. Their teams refuse to use it. This isn't a training problem, or a change management problem. It's a coherence problem: incentives, processes and metrics still reward the very behaviours AI was meant to replace.
Analysis
L'avènement de l'IA agentique rend obsolète l'immense infrastructure méthodologique bâtie par l'industrie Tech depuis vingt ans. Alors que la production de code devient une commodité, le véritable défi des dirigeants n'est plus technologique, mais structurel : il s'agit de démanteler l'encombrant "exosquelette Agile" devenu un frein, pour réallouer massivement le capital financier et humain de l'usine à code vers la stratégie et la découverte de valeur.