Anthropic Acquires Vercept to Deepen Claude’s Computer-Use AI Capabilities
Seattle is not just shipping coffee and cloud credits right now. It is exporting agents. Vercept emerged from the AI2 incubator with Vy, a computer-use agent that could see a remote Mac, interpret the interface, and execute workflows from natural-language prompts. No API dependency. No brittle integrations. Just vision layered over software, moving through tasks the way a human would, except without the coffee breaks. In a cycle where every founder claims automation, Vercept actually automated the act of using the computer itself.
On February 24, 2026, Anthropic made it official: it acquired Vercept to strengthen Claude’s computer-use capabilities. By February 25, 2026, the broader media cycle amplified the signal. This is the kind of tech news that lands quietly but carries weight. Not because of theatrics, but because of direction. Claude is not just answering questions anymore. Claude is learning to operate.
The team matters. Kiana Ehsani, CEO and co-founder, is joining Anthropic alongside co-founders Luca Weihs and Ross Girshick. Vercept raised $50M in total, including a previously disclosed $16M seed led by Fifty Years with Point Nine, AI2 Incubator, and angels Eric Schmidt, Jeff Dean, Kyle Vogt, and Arash Ferdowsi. For a company just over 1 year old in early 2026, that is velocity. Capital followed conviction. Talent followed capability.
There is context beneath the headline. Matt Deitke, Vercept co-founder, departed in 2025 for Meta’s Superintelligence Lab on a reported $250M package. That move signaled how intense the agent arms race has become. The real currency is not capital. It is engineers who can make models act. In this layer of tech news, acquisitions are less about buying revenue and more about compressing years of iteration into quarters.
Customers now face a hard transition. Vercept’s product will shut down on March 25, 2026, with roughly 30 days to migrate. Vy as a standalone service ends. What survives is the capability. This is the trade founders make when speed outruns independence. You give up surface area and gain distribution.
Anthropic has not published a roadmap for when Vercept-derived functionality lands inside Claude. What it has published is intent. Claude will move deeper into direct computer interaction. That shifts the competitive posture. Models that merely respond will feel thin next to models that execute. The acquisition reads less like expansion and more like alignment with where enterprise AI is heading.
This is not funding theater. It is structural positioning inside the AI stack. In today’s tech news, the winners are not the loudest labs. They are the ones quietly assembling the teams that can turn reasoning into action.
If perception was Vercept’s edge, and scale is Anthropic’s advantage, the real question is not whether Claude can use a computer. It is how fast it becomes the default way work gets done when no one is watching the mouse move.