Back to Blog
EnterpriseAI AgentsAnthropic

Anthropic Just Dropped Claude Cowork and Acquired Vercept — Here's What It Means for Every Business

Mark Cijo·

In the last week of February 2026, Anthropic made two moves that sent shockwaves through the enterprise software industry. First, they expanded Claude Cowork with enterprise plugins that connect Claude directly to Google Workspace, Slack, DocuSign, FactSet, LegalZoom, and dozens more business tools. Then, the very next day, they acquired Vercept — a Seattle-based AI startup that specializes in computer-use agents capable of operating desktop software like a human would.

The software industry's response was immediate and dramatic. Adobe, Salesforce, and ServiceNow all saw their shares slide 25 to 30 percent in the selloff that followed. CNBC called it the "dark side of AI." Wall Street called it a generational disruption event.

I run 18 AI agents in production across my operations. When I saw what Anthropic shipped, my reaction was not panic — it was recognition. This is exactly the direction I have been building toward since launching my 18-agent system in January 2026. The difference is that now the largest AI companies in the world are validating the model at enterprise scale.

What Claude Cowork Actually Is

Claude Cowork is Anthropic's enterprise AI product that turns Claude from a chat interface into a workspace teammate. It launched in late January 2026 and has been expanding rapidly since then.

Here is what makes Cowork different from just using Claude in a browser:

Enterprise plugins. Claude can now connect directly to the tools your business already uses. The current plugin ecosystem includes Google Workspace (Calendar, Drive, Gmail), DocuSign, Apollo, Clay, Outreach, Similarweb, MSCI, LegalZoom, FactSet, WordPress, Harvey, Slack, LSEG, S&P Global, Common Room, and Tribe AI. More are being added constantly.

Department-specific agents. Plugins turn Claude into specialized agents for every department. HR gets resume screening and onboarding automation. Investment banking gets financial modeling and deal analysis. Design gets creative asset management. Engineering, operations, private equity, and wealth management all get purpose-built capabilities.

Cross-application orchestration. Claude can now work across Excel and PowerPoint end-to-end, passing context between applications. It does not just help you with one task in one app — it moves between tools the way a human colleague would, carrying context from one step to the next.

Private plugin marketplaces. Enterprises can build and distribute custom plugins across their organization through private marketplaces. This means a company can create internal Claude plugins tailored to their specific workflows and deploy them to every team.

Why This Is Different

Claude Cowork is not a chatbot that summarizes documents. It is an AI agent that connects to your business tools, operates across them, and executes multi-step workflows. The plugin architecture means it can expand to cover virtually any enterprise software stack.

What the Vercept Acquisition Means

The day after expanding Cowork's enterprise plugins, Anthropic announced the acquisition of Vercept — a Seattle AI startup that had raised $50 million and built sophisticated computer-use agent technology.

Vercept's core product was an AI agent that could complete tasks inside desktop applications the same way a human with a laptop would. Click buttons, fill forms, navigate menus, read screens, and execute multi-step workflows across any software interface.

The acquisition brought Vercept's co-founders and team into Anthropic: CEO Kiana Ehsani, Luca Weihs, and Ross Girshick (a former Meta AI researcher). One of Vercept's founders had been recruited by Meta before the acquisition, which gives you an idea of the caliber of talent involved.

Why this matters: Claude already had computer-use capabilities in beta. The Vercept acquisition is Anthropic accelerating that capability — hard. By bringing in a team that has spent years building production-grade computer-use agents, Anthropic is signaling that they want Claude to be able to operate any software, not just software with a plugin API.

The combination is powerful:

  • Plugins give Claude structured, high-performance access to major enterprise tools
  • Computer-use gives Claude the ability to operate any software that does not have a plugin, by literally using it the way a human would

Together, this covers virtually every piece of software a business uses.

Why Software Stocks Crashed

The market reaction to Claude Cowork was severe. Software companies lost billions in market value in the days following the launch. Here is why:

The threat is existential for some categories. When an AI agent can operate legal document software, CRM systems, financial analysis tools, and HR platforms autonomously, the value proposition of many SaaS products shifts. The software becomes the substrate, but the user becomes the AI — not a human paying $50/month per seat.

The partner list paradox. After the initial selloff, companies named as Anthropic partners and integration targets — Salesforce, DocuSign, LegalZoom, Thomson Reuters, FactSet — all rallied sharply. The market realized that being integrated with Claude Cowork is better than being replaced by it. Companies that plugged in became more valuable. Companies that did not became more vulnerable.

The long-term implication. Software companies that build for AI agents as first-class users of their products will thrive. Software companies that only build for human users will face margin compression as AI agents handle more of the work that humans currently do inside their products.

This is not theoretical. It is happening right now, and the stock market is pricing it in.

What This Means for Businesses

If You Are Deploying AI Agents

Claude Cowork validates the multi-agent model I have been building for clients. The plugins create a standardized way for AI agents to access enterprise tools — which is exactly what MCP (Model Context Protocol) does at the protocol level.

The difference between what I build and what Cowork offers out of the box:

  • Cowork gives you a single AI agent (Claude) connected to multiple tools via plugins. It is powerful for individual productivity and team workflows.
  • A multi-agent system (what I build with OpenClaw) gives you specialized agents for each function, coordinating through an AI COO, with role-specific permissions, audit trails, and business logic. It is designed for operations at scale.

Both approaches are valid. Cowork is excellent for teams that want AI-enhanced productivity within their existing workflows. Multi-agent systems are for businesses that want AI to own and operate entire departments autonomously.

If You Use Enterprise Software

Pay attention to which of your software vendors are building Claude plugins and AI agent integrations. The ones that are will become more valuable as AI agents become primary users of their products. The ones that are not will slowly become less relevant.

Ask your vendors: "Does your product support MCP? Do you have a Claude Cowork plugin? Can AI agents interact with your product programmatically?" If the answer is no across the board, start evaluating alternatives.

If You Are in the Software Industry

The Cowork launch is a signal that AI companies are coming for enterprise workflows — not just productivity features. Building an AI layer on top of your product is no longer a nice-to-have. It is existential. The companies that survived the February selloff are the ones that had already built AI integrations. The ones that got hammered are the ones that had not.

The Acceleration Is Real

In February 2026 alone: Anthropic launched Cowork with enterprise plugins and acquired Vercept. Atlassian launched AI agents in Jira. NIST launched AI agent standards. OpenAI raised $110 billion. The infrastructure for AI agent workforces is being built at every level simultaneously.

What I Am Doing About It

I am evaluating the Cowork plugin architecture for integration with the multi-agent systems I build for clients. The potential is clear:

  1. Use Cowork plugins as tool connectors within specialized agents, giving each agent native access to enterprise software
  2. Leverage computer-use capabilities for tools that do not have plugins yet, bridging the gap until the ecosystem matures
  3. Build custom plugins for clients' internal tools and distribute them through private marketplaces

The Anthropic ecosystem is becoming a credible foundation for enterprise AI agent deployment. Combined with MCP for protocol-level integration and A2A for agent-to-agent coordination, the stack is coming together faster than most people realize.

The Bottom Line

Anthropic is not building a chatbot. They are building an AI workforce platform. Cowork gives Claude access to enterprise tools. Vercept gives Claude the ability to operate any software. Together, they create an AI agent that can do real work across real business systems.

The software industry is repricing around this reality. Businesses that deploy AI agents now — whether through Cowork, custom multi-agent systems, or both — will have a structural advantage that compounds over time. The tools are here. The integrations are live. The question is how fast you move.

If you want to understand how Claude Cowork, multi-agent systems, and the broader AI agent ecosystem fit into your specific operations, book a discovery call. I will map out exactly where agents create the most value for your business.

Want an AI Workforce for Your Business?

Book a free call and I'll show you exactly where AI agents fit in your operations.

Book a Free Call

Enjoyed this post?

Get notified when I publish new insights on AI agent systems.

By subscribing, you agree to our Privacy Policy. Unsubscribe anytime.