Claude 3.5 Sonnet: Anthropic Is Costing Wall Street Sleep and Billions of Dollars

Imagine you spent years building a multibillion-dollar empire by selling a convenient interface on top of a database. Then one company releases code that no longer needs your interface. That is exactly what happened last night, when Anthropic unveiled an updated Claude 3.5 Sonnet.

The key point is not that it has gotten better at writing poems or code. It has learned to use a computer literally like a human: look at the screen, move the cursor, click buttons and type text in any application.

Wall Street felt the moment instantly. Shares of companies whose business is built on providing software tools to other businesses slid. Investors grasped a simple truth: if an AI agent can log into an old CRM system on its own, move data from Excel and send an email in Outlook, the value of "pretty buttons" and "convenient dashboards" tends toward zero. A huge segment of the SaaS market is turning into a set of unnecessary middlemen between AI and the result.

Why does this matter right now? Until this moment we lived in the "Copilot" paradigm, where a neural network sits in the corner of the screen and offers advice. Anthropic skipped that stage and proposed the concept of an autonomous agent. It needs no special APIs or complex integrations that companies spent years configuring. Claude simply "sees" the pixels on the screen and understands what to do. It is an elegant and, at the same time, unsettling solution that sidesteps all the classic barriers to adopting new technology.

Recall how OpenAI promised us something similar with its "operators," yet Anthropic was the first to release a working solution publicly. It is a classic example of how the number two player in a market can seize the initiative simply by giving users a tool for real work rather than yet another demo of the future.

The problem is that this future has arrived too fast for the stock market, which is used to valuing software giants by the number of licenses sold. Now a single Claude license can replace dozens of workstations equipped with complex software.

The consequences for the industry will be tectonic. Where developers once spent months building integrations between different services, an agent now does it "with its eyes." That means the death of many middleman startups and a serious challenge for giants like Salesforce or HubSpot. They will have to either reinvent themselves completely or watch their market capitalization evaporate under the onslaught of agents that do not need their interfaces. Anthropic has effectively declared war on traditional software, and the first losses in that war already run into the billions of dollars.

Of course, the technology is not perfect yet. Claude can make mistakes, miss buttons or get stuck on complex CAPTCHAs. But the direction is set very clearly. We are moving from the era of "software for people" to the era of "infrastructure for agents." And those who lost money on the stock market today may have been the most farsighted, realizing in time that the rules of the game had changed forever. While others debate chatbot hallucinations, Claude is already learning to close your tasks in Jira.

The bottom line: Anthropic has set a precedent in which an AI model becomes not just a tool but a full-fledged user. Will OpenAI be able to respond with something more significant than another voice demo?