Tuesday, February 3, 2026 will go down in history as the day when agentic AI stopped being a bet for the future and became a real elephant in a china shop. Anthropic launched its Legal plugin for Claude Cowork: complete automation of contract review, NDA management, compliance flows and legal reports without human intervention. The stock market result was immediate: $285 billion evaporated on Wall Street before the opening. The dream of any anti-establishment nerd: an epoch-making earthquake caused by publishing code on GitHub.
But what happened to cause such a catastrophe? The key is to look towards the future and how processes and tools from the past have remained dormant for too long in sectors with great inertia towards technological change. Investors, who have more work every day interpreting the tsunamis caused by each announcement related to AI, understood in minutes that something is changing drastically and it’s not for tomorrow, it’s for today.
The crash that everyone should study
Thomson Reuters fell 16%, its worst day in history. RELX (LexisNexis) lost 14%, Wolters Kluwer 13%. The financial earthquake did not stop there, it reached the heart of the industries based on data arbitration and identity verification. Giants like Experian, Equifax y TransUnion experienced sessions of extreme volatility amid fears that agentic AI will democratize risk analysis and validation of credit profiless, functions that until today justified their high margins.
As happened with Thomson Reuters and RELX, the markets began to discount that the value no longer lies in possessing the data (increasingly more accessible and processable by models such as Claude or GPT) but in the ability to execute autonomous actions with it. If an AI agent can audit flows of compliance and solvency in seconds through the Anthropic plugin, the consultation income model of these agencies faces inevitable obsolescenceconsolidating a collapse that reflects the end of the era of static information intermediaries.
The vulnerability of these intermediaries lies in the fact that their business model is not based on the creation of value, but on the exploitation of information asymmetry. Like credit rating agencies, these companies have thrived by acting as “guardians of the truth” in markets where data was fragmented, but agentic AI has destroyed that defensive moat by enabling real-time audits and autonomous verifications that turn the static report into a documentary fossil.
The coming obsolescence
This trajectory of obsolescence imminently threatens market intelligence consultancies whose sector analyzes They can’t compete with real-time sentiment monitoringto notary and registration agencies that act as mere identity validators, and to insurance brokers that operate as simple information gateways.
Likewise, the collapse will affect B2B prospecting list providers and background verification companies, since an agent can locate decision-makers or validate qualifications in seconds without depending on closed databases. Finally, media and programmatic purchasing centers will see their intermediation margin disappear in the face of an AI capable of executing direct purchases based on immediate return on investment. consolidating a scenario where the market no longer pays for access to databut by the ability to execute actions on it.
But What is truly significant is what has happened outside the scope of the “data guardians”: Adobe fell to six-year lows, PayPal fell 20%, Gartner another 20%. The iShares technology software index lost 5.7% in a single session. Why did companies with no apparent relationship with the legal sector fail the same? Because investors did the math that many CIOs have been avoiding for months: the SaaS model based on seats per user is going to be seriously compromised. The idea that investors have envisioned is that agents can not only potentially replace employees but that they will not need the licenses they use.
The most important thing about all this (besides the tremendous downturn in the stock market) is what the actors of this umpteenth revolution caused by AI do from now on. Faced with this paradigm shift, the CIO must abandon the role of license manager to become an efficiency architectauditing its technological stack to transition from pay-per-seat models to models based on consumption or resultswhile ensuring robust data governance that enables agent integration without compromising security.
For their part, those responsible for the “data guardian” companies need urgently pivot from being mere information stores to becoming execution platformswhere the value does not reside in the consultation of static data, but in the ability to offer layers of agentic intelligence that solve business problems in real time. Possession of the data currently gives them the possibility of developing these services with an advantage, but that advantage will quickly fade.
Finally, SaaS software multinationals must reinvent their revenue structure before the market penalizes them, migrating from monetization of workforce (seats) to monetization of value generated by automationintegrating native agentic capabilities that justify their permanence compared to customized solutions developed internally with AI.
I know that they are reflections of radical and profound changes in not one but several very important economic sectors, all caused by the rise of new AI tools. It is not unreasonable to think that the earthquakes have not ended. Not at all…
In fact, a few hours after finishing this article, OpenAi announced Frontier, a platform designed to orchestrate fleets of autonomous agents that operate directly on giants like Salesforce and Workday. So the agents They threaten to claim a new victim with CRMs and HR platforms.
