The smartest chatbot in the world is still just a chatbot. Anthropic knows this — and it’s spending aggressively to change it.

On Wednesday, Anthropic acquired Vercept, a Seattle-based startup that built AI agents capable of remotely operating a full MacBook. It’s the company’s second major acqui-hire in three months, following its December purchase of Bun (the engine behind Claude Code). The message is clear: Anthropic isn’t building a better conversation partner. It’s building a digital employee.

What Vercept Actually Built

Vercept came out of AI2, Paul Allen’s prestigious AI incubator. Its product, Vy, was a cloud-based agent that could take control of a remote Mac and complete complex tasks autonomously — opening apps, clicking menus, copying data between tools, filling spreadsheets. Not generating text about how to do work. Actually doing it.

The startup raised $50 million, with an angel roster that included former Google CEO Eric Schmidt, DeepMind chief scientist Jeff Dean, and Cruise founder Kyle Vogt. Serious money from serious people who believed the thesis.

After just over a year, Vercept is shutting down its standalone product by March 25. CEO Kiana Ehsani, CTO Luca Weihs, and researcher Ross Girshick are all joining Anthropic to supercharge Claude’s computer use capabilities.

The $250 Million Defection and the LinkedIn Meltdown

Not everyone made the jump. Co-founder Matt Deitke negotiated a reported $250 million compensation package to join Meta’s Superintelligence Lab instead. A quarter of a billion dollars for one researcher. Welcome to the NBA economy of frontier AI.

Meanwhile, Vercept co-founder and investor Oren Etzioni — the legendary founding CEO of the Allen Institute for AI — went public with his frustration. “After a little bit more than a year, Vercept is throwing in the towel,” he wrote on LinkedIn, calling the 30-day customer migration window “sad.”

Lead investor Seth Bannon fired back: “You disparaged the heroic work of the founders for achieving an outcome most could only dream of.”

Entertaining Silicon Valley theater. But it reveals the extraordinary pressure cooker AI startups operate in right now. When your co-founder gets poached for a quarter-billion and your board is fighting on LinkedIn, “normal market” doesn’t apply.

Anthropic’s Acquisition Playbook

Zoom out and a pattern emerges. Anthropic is systematically assembling the full stack of agent capabilities:

  • Code execution — Bun acquisition → Claude Code
  • Computer use — Vercept acquisition → desktop automation
  • Enterprise integration — Claude plugins for Excel, PowerPoint, HR systems (announced last week)

This isn’t a company that thinks the future is chatbots with better personality. It’s a company that thinks AI should do your job’s boring parts while you do the interesting ones. And it’s acquiring rather than building from scratch, which tells you something about urgency.

As Gartner analyst Ashish Banerjee put it: “In frontier AI, talent retention is the new uptime. If a provider can’t keep its builders, it can’t keep its roadmap.”

What Computer Use Actually Looks Like

“Computer use” sounds abstract until you picture it concretely.

Imagine telling Claude: “Find the cheapest flight to Tokyo for the last week of March, book it with my miles, and add it to my calendar.” Today’s AI can tell you how to do that. A computer-use agent would do it — opening your browser, navigating to the airline, logging in, comparing options, completing the booking, updating your calendar. While you get coffee.

Scale that to enterprise. Two hundred onboarding forms. Financial reconciliation across three systems. Marketing campaigns across a dozen platforms. Hours become minutes.

This is why analysts see the Vercept deal as part of an inevitable consolidation wave. The long-term path for computer-use tech runs through integration into larger platforms with scale, data access, and tight model alignment — not standalone startups.

The Warning for Early Adopters

There’s a cautionary tale here. Vercept’s enterprise customers just got 30 days’ notice to migrate off the platform. If you built workflows on Vy, you’re scrambling.

This is the classic startup risk, amplified by AI’s blistering pace of consolidation. Your vendor could get acqui-hired into a larger company at any moment. The smart move, per Omdia chief analyst Lian Jye Su: “Build in risk mitigation — low-commitment experiments, data portability requirements, modular architecture using APIs and open standards.”

Don’t bet your operations on an AI startup without an exit plan.

Where This Heads

We’re entering the era where AI stops being something you talk to and becomes something that works for you. Anthropic, OpenAI, Google, Meta — they’re all converging on the same vision: agents that operate software, complete multi-step workflows, and handle the tedious digital labor eating most knowledge workers’ days.

With Vercept’s team now building for Claude, expect computer use to improve fast through 2026. The question isn’t whether AI agents will use your computer as well as you do. It’s whether you’ll trust them to — and how quickly your employer decides that answer is yes.