In my time leading product at Vestra AI, we saw over 1,500 AI agents created on our no-code platform. The single biggest lesson from that experience was this: the most successful agents weren’t just chatbots with fancy prompts; they were deeply integrated into workflows, capable of interacting with tools and data to perform meaningful work. This is the essence of high agency—the ability to not just answer questions, but to take action and solve problems autonomously.
That’s why Anthropic’s recent launch of Cowork plugins for Claude isn’t just another feature announcement. It’s a tectonic shift. When the news dropped that their new legal plugin had wiped billions off the market value of established legal tech companies, it wasn’t an anomaly; it was a tremor signaling a much larger earthquake. [1] We are witnessing the unbundling of specialized software, as capabilities that once required dedicated applications are now being absorbed into foundational AI platforms as simple, installable plugins.
Most people are focusing on the individual plugins—the legal assistant, the finance tool, the marketing creator. They’re missing the bigger picture. The real story isn’t the plugins themselves, but the open-source infrastructure they’re built on: the Model Context Protocol (MCP). [2] This is the quiet revolution that will dismantle and rebuild the software industry as we know it. In this article, I’ll break down the three-fold impact of this shift: the immediate disruption to the software industry, the new high-agency playbook for professionals, and the stark economic realities facing startups and enterprises alike.
The Great Unbundling: How Cowork and MCP Are Rewriting the Rules of Software
For years, the software-as-a-service (SaaS) model has been defined by specialized, vertical applications. You had a separate tool for everything: one for legal contract management, another for financial analysis, and yet another for marketing content creation. Anthropic’s Cowork plugins are a direct assault on this paradigm. The immediate, brutal market reaction to their legal plugin—which sent shares of established players like RELX (LexisNexis), Thomson Reuters (Westlaw), and LegalZoom tumbling—was not an overreaction. [1] It was a rational pricing-in of a fundamental shift in the value chain.
The iShares Expanded Tech Software ETF’s dip of over 2% following the announcement was a clear signal that this isn’t about one company or one plugin. [1] It’s about the entire software ecosystem. Why would a startup pay for a standalone contract review tool when they can get a “good enough” (and rapidly improving) version as a simple plugin for the AI they already use for a dozen other tasks? The barrier to entry for many software categories has just been obliterated.
However, the true disruptive force here is not the plugins themselves, but the open-source Model Context Protocol (MCP) they are built upon. Anthropic’s decision to open-source MCP in November 2024 and then donate it to the Linux Foundation in December 2025 was a masterstroke of product strategy. [2] [3] By creating a universal standard for AI agents to connect with external data and tools, they are fostering an open ecosystem, not a walled garden. This is the playbook that made web browsers and mobile operating systems the dominant platforms of their time. It invites collaboration and innovation from the entire community, as evidenced by early adopters like Block, Apollo, and Sourcegraph who are already building on MCP. [2]
As Block’s CTO, Dhanji R. Prasanna, put it, “Open technologies like the Model Context Protocol are the bridges that connect AI to real-world applications, ensuring innovation is accessible, transparent, and rooted in collaboration.” [2] This isn’t just about making Claude more powerful; it’s about creating the foundational layer for the next generation of software—one where value is created not by building standalone apps, but by building specialized skills and data connectors that can be seamlessly integrated into any agentic AI.
The High-Agency Professional: A New Playbook for the AI Era
While the industry-level disruption is profound, the more immediate impact will be felt at the individual level. The rise of agentic AI and tools like Cowork plugins fundamentally changes the playbook for knowledge workers. The fear of job displacement is palpable—employee concerns about being replaced by AI have jumped from 28% in 2024 to 40% in 2026, and a majority of executives (54%) expect AI to displace jobs. [4] [5]
However, I believe this focus on replacement is misguided. The real story is augmentation. As I’ve seen firsthand in scaling products to millions of users, technology rarely causes mass unemployment overnight. Instead, it creates a performance gap between those who adopt it and those who don’t. The professionals who thrive in this new era will be those who exhibit high agency—proactively using these tools to amplify their capabilities, automate mundane tasks, and focus on high-leverage strategic work.
Consider the practical applications of the initial Cowork plugins:
| Department | Traditional Workflow | High-Agency Workflow with Cowork |
|---|---|---|
| Product Management | Manually synthesizing user feedback from various sources to write a feature spec. | Use the Product Management plugin to automatically ingest user research from documents and generate a first-draft spec, freeing up the PM to focus on strategy and validation. |
| Marketing | Spending hours analyzing campaign data in spreadsheets to create a performance report. | Use the Marketing plugin to connect directly to analytics platforms, generate a report with key insights, and even draft initial content for the next campaign. |
| Legal | Manually reviewing a standard NDA, a time-consuming task for a small legal team. | Use the Legal plugin to perform an initial review, flag non-standard clauses, and ensure compliance, reducing review time from hours to minutes. |
This isn’t about replacing the product manager, marketer, or lawyer. It’s about transforming their roles. It’s about automating the 80% of their work that is repetitive and low-value, so they can dedicate their expertise to the 20% that truly matters. This is the democratization of expertise. A startup founder without a legal team can now get a first-pass contract review. A junior analyst can generate data visualizations that once required a data science degree. This is a massive unlock for productivity and innovation.
From my experience at ShareChat, the speed of iteration is everything. Tools like Cowork plugins are the ultimate iteration accelerators. They compress the time it takes to go from idea to execution, allowing teams to test more hypotheses, learn faster, and ultimately build better products.
The Economic Realities: A Warning and an Opportunity
The economic implications of this shift are stark. The AI agents market is not a niche; it’s a juggernaut projected to grow from approximately $7.8 billion in 2025 to over $52 billion by 2030, a staggering 46.3% compound annual growth rate. [6] This isn’t a distant future; it’s happening now. The productivity gains are real, with Anthropic’s own research suggesting a potential 1.8 percentage point increase in US labor productivity growth from widespread AI adoption. [7]
For the startup ecosystem, this is both a warning and an opportunity. The warning is clear: if your startup’s value proposition is simply a thin wrapper around a task that can be automated by a Cowork plugin, you are on borrowed time. The launch of Cowork, reportedly built in just a week and a half using Anthropic’s own tools, demonstrates how quickly foundational AI companies can bundle and deploy features that were once entire businesses. [8]
This is the harsh reality of building on someone else’s platform. The startups that survive and thrive will be those with a defensible moat that goes beyond simple automation. This could be proprietary data that a general-purpose AI cannot access, deep domain expertise that informs a highly specialized workflow, or a superior, user-centric experience that a generic plugin cannot match. The focus must shift from what you do to how you do it and what unique data you bring to the table.
For enterprises, the opportunity is immense. The 66% of organizations already reporting productivity gains from AI are just scratching the surface. [9] The companies that will win are those that move beyond ad-hoc AI experiments and develop an enterprise-wide strategy for agentic AI adoption. This means identifying key workflows, investing in training, and building custom plugins and MCP connectors to integrate AI deeply into their core operations. The future of competitive advantage will be measured by how effectively an organization can augment its human talent with a workforce of AI agents.
Conclusion: The Future is Agentic—and It’s Open Source
The launch of Cowork plugins is more than just a new feature; it’s a declaration of a new era. The paradigm of standalone, specialized software is giving way to a world of intelligent, agentic platforms extensible through an open ecosystem of plugins. The core product is no longer a single-purpose application but the underlying AI agent and the protocol that connects it to the world.
As product managers, engineers, and business leaders, we have a choice. We can view this as a threat, clinging to old models as they become obsolete. Or we can embrace it as the single greatest opportunity of our careers. The opportunity to build more, faster, and with greater impact than ever before. The opportunity to offload the mundane and focus on the meaningful. The opportunity to exhibit high agency.
My advice is simple: start building. Don’t wait for a perfect, risk-free version. Install the Claude Desktop app, experiment with the existing Cowork plugins, and, most importantly, start thinking about what custom plugins you can create for your own workflows. The tools are here, they are accessible, and they are powerful. The future won’t be built by those who wait; it will be built by those who take ownership and start shipping.
References
[1] Sherwood News. (2026, February 3). Anthropic’s legal plug-in for Claude Cowork prompts rush out of legal software and publishing stocks. https://sherwood.news/markets/anthropics-legal-plugins-for-claude-cowork-prompt-rush-out-of-legal-software/
[2] Anthropic. (2024, November 25). Introducing the Model Context Protocol. https://www.anthropic.com/news/model-context-protocol
[3] Anthropic. (2025, December 9). Donating the Model Context Protocol and establishing the Agentic AI Foundation. https://www.anthropic.com/news/donating-the-model-context-protocol-and-establishing-of-the-agentic-ai-foundation
[4] CNBC. (2026, January 20). AI impacting labor market ‘like a tsunami’ as layoff fears mount. https://www.cnbc.com/2026/01/20/ai-impacting-labor-market-like-a-tsunami-as-layoff-fears-mount.html
[5] World Economic Forum. (2026, January 7). Four Futures for Jobs in the New Economy: AI and Talent in 2030. https://reports.weforum.org/docs/WEF_Four_Futures_for_Jobs_in_the_New_Economy_AI_and_Talent_in_2030_2025.pdf
[6] MarketsandMarkets. (n.d.). AI Agents Market Size, Share, Growth & Latest Trends. https://www.marketsandmarkets.com/Market-Reports/ai-agents-market-15761548.html
[7] Anthropic. (2026, January 15). New building blocks for understanding AI use. https://www.anthropic.com/research/economic-index-primitives
[8] Fortune. (2026, January 13). Anthropic launches Cowork, a file-managing AI agent that could threaten dozens of startups. https://fortune.com/2026/01/13/anthropic-claude-cowork-ai-agent-file-managing-threaten-startups/
[9] Deloitte. (n.d.). The State of AI in the Enterprise - 2026 AI report. https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html