--- title: "Why MCP Is The Doorway To API-Based Business" description: "When we talk about MCP, we're really talking about APIs — and the vast economy behind monetized API products." canonicalUrl: "https://zuplo.com/blog/2026/03/25/why-mcp-is-the-doorway-to-api-based-business" pageType: "blog" date: "2026-03-25" authors: "billDoerrfeld" tags: "Model Context Protocol, API Monetization" image: "https://zuplo.com/og?text=Why%20MCP%20Is%20The%20Doorway%20To%20API-Based%20Business" --- The tech industry is still rallying strongly behind MCP, with more and more AI frameworks supporting it and MCP servers emerging for [DevOps tooling](https://www.infoworld.com/article/4096223/10-mcp-servers-for-devops.html) and [major clouds](https://www.infoworld.com/article/4129024/five-mcp-servers-to-rule-the-cloud.html). But many still overlook a simple truth: MCP is almost always a doorway to an underlying API. Zuplo's 2025 [State of MCP report](https://zuplo.com/mcp-report) found that 58% of MCP builders wrap existing APIs with MCP. Additionally, 56% of MCP servers are developed for external customers, too, meaning many sit directly in front of APIs exposed through digital products and consumer-facing applications. Given that 65% of organizations generate revenue from their APIs, according to Postman's 2025 [State of The API Report](https://www.postman.com/state-of-api/2025/), MCP introduces clear opportunities for both direct monetization and new business models. In this world, MCP is becoming the new interface for the agent economy. It's the API layer for business workflows triggered by natural language. Below, we'll look at some emerging case studies that demonstrate how MCP is delivering an ROI, as well as tips on how to get the most out of it for API-based business. ### Key Takeaways - **AI chat is becoming a business channel**: Consumer brands like Walmart and Expedia are experimenting with agentic commerce, while OpenAI is investing in ads and monetization features. - **The industry is primed for API monetization**: MCP is set to accelerate AI agent-driven business through more monetized, API-based interactions. - **MCP enables business-user success**: Enterprise MCP features are already empowering workflows across platforms like Workato, Square, and others at scale. - **Investment in MCP is growing**: With over 18K servers and support from developer tools and major cloud providers, MCP adoption is accelerating. - **Getting MCP right requires more than technology**: It demands strong product thinking, along with excellent documentation and API design for discoverability. ## MCP Is An Agentic Commerce Enabler MCP can be embedded into chatbot environments to support monetized actions such as product orders, service purchases, and bookings, all within an AI conversation. While not all current AI chat integrations rely on MCP, it's primed to be an experience layer for more autonomous AI. By introducing MCP into AI chats, businesses can capture intent at the source, leading to higher conversion rates. Shopify is a strong example of this. Its [Storefront MCP server](https://shopify.dev/docs/apps/build/storefront-mcp) enables AI agents to connect directly to product catalog, checkout, and fulfillment APIs, creating a new revenue channel for merchants without requiring a traditional storefront experience. Other consumer brands are experimenting with agentic commerce. For example, Walmart has built [dozens of AI agents](https://www.wsj.com/articles/why-walmart-is-overhauling-its-approach-to-ai-agents-4b1fbc65) and is using MCP behind the scenes to connect them, advancing supply chain and inventory management. Walmart's recent integration with ChatGPT allows users to search and purchase items, from groceries to electronics, directly within chat. Travel is another clear example. Take a truly unified booking workflow: It will require API access for an agent to synchronize flights, hotels, pricing, and rental cars to create a streamlined booking experience. Expedia is already inching toward agentic commerce — currently [integrating with ChatGPT](https://www.expedia.com/product/expedia-in-chatgpt/) using ChatGPT Connectors to retrieve data and link to customized bookings on an external page. MCP could take this flow further to support a more unified, agent-driven orchestration layer over time. Fintech platforms are moving in a similar direction. [Monetized financial APIs](https://zuplo.com/blog/5-api-monetization-success-stories) are already offering MCP servers to extend their platforms. For instance, AI agent platform Dust is [using Stripe's MCP](https://stripe.com/en-hu/customers/dust) to access payment data and streamline actions like refund automation, while Plaid is developing [similar tooling](https://plaid.com/docs/resources/mcp/). With MCP supporting financial transactions, agentic commerce is poised for significant change. An agent can theoretically initiate payments, manage subscriptions, or trigger billing flows. Although it will require additional protocols to realize agentic commerce, MCP is likely to play a central role in standardizing how agents invoke these capabilities. ## The Significant Business Effects of MCP In some cases, MCP server usage directly generates revenue, especially for those servers connected to metered, usage-based APIs with pay-as-you-go pricing. But beyond direct monetization, MCP servers also unlock a range of indirect business benefits, from amplifying a developer platform to transforming enterprise business workflows. ### MCP Benefits Developer Platforms One of the most immediate gains is developer experience. By extending existing platforms into agent workflows, MCP enables more seamless interaction with complex systems. For example, [Multiplayer](https://leaddev.com/ai/lessons-learned-launching-mcp-server), a full-stack debugging tool, has shown how its MCP server helps its customers automatically ingest session data, giving AI coding agents real-time context to act on. MCP also simplifies how developers interact with platforms by providing a natural language option to trigger complex underlying commands. GitHub's MCP server, for instance, allows maintainers to classify and group contributions using plain English, reducing cognitive load and increasing platform stickiness. ### MCP Supercharges Enterprise Business But the benefits go well beyond individual developer productivity — MCP enables businesses to participate more directly in the agentic AI economy. Take [the case of Workato](https://www.workato.com/the-connector/ai-usage-growth/), which noticed Claude usage soar by 700% when they gave their 1,000-person workforce access to Enterprise MCP due to its ability to make agents far more actionable with business apps. With MCP, Workato employees can [orchestrate business processes](https://leaddev.com/ai/how-to-justify-ai-investments) across applications, connecting CRMs, responding to support tickets faster, empowering sales teams, and reducing churn. These workflows rely on API connectivity under the hood, which MCP helps operationalize at scale and in real time. Block has seen [similar success](https://thenewstack.io/how-block-got-12000-employees-using-ai-agents-in-two-months/) by enabling MCP access for its [Goose](https://block.github.io/goose/) agent across internal teams. At a broader level, MCP servers that wrap APIs for business platforms like Salesforce and HubSpot allow agents to create records, trigger campaigns, and update pipelines based on real-time context. The result is more responsive systems, better lead enrichment, and smarter, automated business processes. In short, MCP doesn't just improve how APIs are used, it reimagines the value of these interfaces entirely, making them increasingly ingrained into how enterprises do business. ## What Most MCP Builders Get Wrong In an agentic world, the agent chooses what tools to use at runtime. Yet many MCP servers aren't treated as the products they are, which limits discovery, utility, and performance in practice. For instance, a common issue is weak documentation. Many MCP servers and their tool descriptions are represented as little more than a [README in a GitHub repository](https://github.com/github/github-mcp-server). While technically sufficient, this falls short of core [API best practices](https://nordicapis.com/best-practices-for-creating-useful-api-documentation/) around having clear, structured, human-navigable documentation that treats the interface as a first-class product. Tool design is another gap. Many MCP servers lack example prompts that show how to query the server or what each tool enables. Without clear, natural language descriptions for tool names and functions, discoverability suffers for both humans and agents. Lastly, security is a frequent weak point. Surprisingly common [security risks](https://thenewstack.io/building-with-mcp-mind-the-security-gaps/) — such as broken authentication, over-permissioning, and misconfigurations — remain widespread throughout MCP servers. These conditions as well as unchecked rate limits can lead to severe security issues and business logic abuse, which prevents a clear [ROI with MCP](https://thenewstack.io/when-is-mcp-actually-worth-it/). ## MCP: Done The Right Way In order to realize MCP as the API business enabler that it truly is, organizations must incorporate more product thinking into their [MCP server design](https://nordicapis.com/8-tips-and-best-practices-for-mcp-server-development/) and deployment practices. Experts share that designing well-scoped MCP servers with more intent is necessary to avoid [runaway token consumption issues](https://thenewstack.io/how-to-reduce-mcp-token-bloat/). In order to position MCP for success as a revenue driver, it will take a blend of technical savvy as well as typical product positioning to treat MCP servers as products. Experts recommend the following techniques for designing and deploying MCP servers: - **Auto-generation**: With MCP generation tools, you can [turn any API into an MCP server](https://zuplo.com/docs/articles/mcp-quickstart). This helps you quickly capitalize on providing MCP while aligning with your platform infrastructure. - **OpenAPI-native**: For MCP servers to match their APIs, one method is to auto-generate MCP servers based on an API definition, such as an OpenAPI specification. - **Safe policies**: Using a tool like [Zuplo for MCP generation](https://zuplo.com/features/mcp-servers) also automatically routes your policies to MCP, meaning no new authorization gaps are present. - **Design for natural language**: MCP must be positioned for agentic consumption. LLMs trained on natural language prompts tend to perform better with tools that are described using simple, everyday language. - **Excellent documentation**: Companies like Figma, GitLab, and PandaDoc realize the benefits of [quality MCP documentation](https://nordicapis.com/5-examples-of-excellent-mcp-server-documentation/), using clear instructions for installation and use, sample natural language prompts, and common troubleshooting questions. - **Use a gateway**: The same API best practice holds true for MCP. [Using a gateway](https://zuplo.com/learning-center/managing-mcp-server-access) is important hygiene to apply proper routing, authentication, authorization, rate limiting, governance, and observability. ## MCP Lets AI Act On The Business World To date, developers have primarily used MCP for context engineering use cases, leveraging servers like Filesystem, GitHub, and Database MCP servers. For instance, [Linear](https://linear.app/), the engineering-focused product management application, has seen over 250,000 users for its MCP server. However, as agentic AI moves past read-only access to become more actionable, it requires deeper API-based access. There's a strong argument that MCP servers represent the future of API-based business as AI begins to take more [real-world actions](https://www.cio.com/article/4018578/why-cios-see-apis-as-vital-for-agentic-ai-success.html) — especially within business applications. As such, MCP is about letting AI act on the world. This could enable immediate in-chat commerce, recurring subscriptions, and autonomous transactions all in a new channel. This new distribution channel essentially expands the reach of an API-driven platform across AI surfaces. Although clear ROI cases are only beginning to emerge, smart companies are realizing [when MCP makes sense](https://thenewstack.io/when-is-mcp-actually-worth-it/). By following these early signals, and building servers that are genuinely useful, discoverable, and secure, MCP builders are well positioned. All in all, it's never been a better time to prepare for the next evolution of API-driven agentic business.