MCP Belgium

Appearance

Use Cases and Real-World Applications ​

Since its introduction by Anthropic in late 2024, MCP has rapidly become the de-facto standard for AI integration. One year after launch, the ecosystem boasts 97M+ monthly SDK downloads, 5,800+ MCP servers, and 300+ MCP clients, with major deployments at Block, Bloomberg, Amazon, and hundreds of Fortune 500 companies. In December 2025, Anthropic donated MCP to the Agentic AI Foundation (AAIF) under the Linux Foundation, co-founded with Block and OpenAI, cementing its status as true industry infrastructure.

This page showcases documented use cases and implementations demonstrating MCP's practical impact.

Developer Tools and IDEs ​

GitHub Copilot & Visual Studio Code ​

Implementation: GitHub's Copilot team integrated MCP support into Visual Studio Code in mid-2025. Developers can configure MCP servers in VS Code settings via an mcp.json file.

Capabilities:

  • Connect to local filesystem servers for project context
  • Access GitHub repositories for issue tracking and PR details
  • Query databases for application data
  • Run tests and deployments through standardized tools

Example Configuration:

json
{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {"GITHUB_PERSONAL_ACCESS_TOKEN": "${GITHUB_TOKEN}"}
    },
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/workspace"]
    }
  }
}

Real-world workflow: Developers can ask AI, "Hey, what changed in this PR?" and the AI calls a GitHub MCP server to fetch PR details, then uses testing tools for comprehensive analysis.

Sourcegraph Cody Integration ​

Challenge: Code AI assistants needed access to project-specific information beyond what was loaded in memory.

Solution: Sourcegraph integrated MCP so Cody could pull data via Sourcegraph's search APIs rather than relying only on in-memory code.

Impact: AI can now search entire codebases when answering questions, dramatically improving the relevance and accuracy of code assistance.

Multi-Platform Adoption ​

Major coding assistant platforms now support MCP:

Windows 11: The "Agentic OS" ​

Announcement: At Microsoft Build 2025, Microsoft unveiled Windows 11 as an "agentic OS" with native MCP support—describing MCP as "USB-C for AI."

Built-in Connectors (December 2025 Insider Preview):

  • File Explorer Connector: AI agents can manage, organize, and retrieve local files with user consent
  • Settings Connector: Natural language control of display, mouse, keyboard, and sound settings on Copilot+ PCs

Security Model: All agent connectors run in secure containers with their own identity and audit trail.

Partnerships: Microsoft is working with Figma, Anthropic, and Perplexity on MCP integrations.

Personal Productivity and Desktop Assistants ​

Claude Desktop and Claude Code ​

Approach: Anthropic's Claude Desktop app uses MCP as the standard mechanism for local system integration.

Available servers:

  • Filesystem server: Read local documents with user permission
  • Calendar integration: Schedule meetings and manage appointments
  • Email composition: Draft messages using actual email clients
  • Note-taking: Connect to personal knowledge management systems

Claude Code 2.0 (2025): The most significant update brought native IDE integration with a VS Code extension, real-time sidebar panel, and autonomous capabilities—Rakuten reported running Claude Opus 4 for 7 hours straight in productive development work.

Cross-platform compatibility: The same MCP servers work with OpenAI's ChatGPT desktop app, Google Gemini CLI, and any MCP-compatible client, demonstrating the protocol's universality.

OpenAI's MCP Adoption ​

March 2025: OpenAI officially adopted MCP across its products, joining the MCP Steering Committee.

Implementation:

  • MCP Connectors in the Responses API with approval workflows for security
  • ChatGPT Desktop supports the same MCP servers as Claude
  • Assistants API deprecation announced (sunset mid-2026), compelling ecosystem migration to MCP

Significance: OpenAI's adoption alongside Google and Microsoft transformed MCP from a vendor-led spec into common infrastructure.

Enterprise Productivity: Block (Square) ​

Implementation: Block (formerly Square) is one of the largest MCP adopters, having developed 60+ internal MCP servers.

Architecture: Integrated MCP with their engineering and project management stack: Snowflake, Jira, Slack, Google Drive, and internal task-specific APIs.

Results: Thousands of Block employees use their AI assistant "Goose" and report cutting up to 75% of time spent on daily engineering tasks.

Key insight: Block's CTO described this as removing "rote work" - the AI handles mechanical tasks across systems, allowing people to focus on creative work.

Enterprise Productivity: Bloomberg ​

Implementation: Bloomberg utilizes MCP internally to help AI developers reduce the time required to ship demos into production.

Architecture: Their engineering team built a system enabling AI agents to interact with the company's entire infrastructure.

Results: Shorter feedback loops and accelerated development cycles for AI-powered features.

Enterprise Knowledge Management ​

Internal Documentation Access ​

Common pattern: Companies build MCP servers for their internal systems:

  • Confluence pages and wikis
  • SharePoint documents
  • Internal APIs and databases
  • Policy manuals and HR data

Security approach: Custom servers implement appropriate access controls, ensuring AI assistants only access data the user is authorized to see.

Business impact: AI assistants can answer questions using actual up-to-date company data rather than generic training information.

Multi-System Integration Example ​

Use case: Customer service representative needs comprehensive customer information.

MCP implementation:

  1. CRM server - Provides customer order history and contact details
  2. Support ticket server - Shows previous issues and resolutions
  3. Knowledge base server - Accesses internal documentation
  4. Communication server - Can send follow-up emails

Workflow: Representative asks, "Show me everything about customer XYZ," and AI coordinates across all servers to provide complete context.

Multi-Tool AI Chains and Orchestration ​

Complex Task Automation ​

Example scenario: "Take the latest sales data, analyze trends, create a slide deck, and email it to the team."

MCP orchestration:

  1. Database tool - Retrieve sales data
  2. Analysis tool - Generate trend insights
  3. Document generator - Create presentation slides
  4. Email tool - Send to team members

Framework integration: Microsoft's Semantic Kernel published guides on using MCP tools as "skills" within AI orchestration pipelines, mixing them with native code capabilities.

Context Continuity Advantage ​

Key benefit: MCP maintains context across tool invocations within the same conversation session.

Practical impact: Outputs from one tool become inputs to the next, with all components connected through a uniform interface, making multi-step workflows more reliable.

Data Analytics and Natural Language Queries ​

AI2SQL Integration ​

Application: Tools like AI2SQL use MCP to connect language models with SQL databases.

Capabilities:

  • List available database tables
  • Execute queries and return results
  • Use sampling feature for natural language to SQL conversion

Example queries:

  • "How many new users joined last week?"
  • "Show me the top-performing products by region"
  • "Generate a sales report for the executive team"

Vector Database and RAG Applications ​

Implementation pattern: Vector databases integrated as MCP resources for retrieval-augmented generation (RAG).

Architecture: Instead of embedding vector search directly in applications, systems treat vector databases as MCP resources where search becomes a tool call.

Advanced capability: IBM noted that using MCP in RAG pipelines allows further tool calls after retrieving information, enabling iterative reasoning over retrieved data.

Academic and Research Applications ​

Zotero Integration ​

Use case: Researchers using MCP servers for Zotero (reference manager) to enhance literature reviews.

Capabilities:

  • Search entire reference libraries semantically
  • Extract annotations and notes from PDFs automatically
  • Generate comprehensive literature reviews by combining multiple sources
  • Track citation relationships and research trends

Multi-source workflows: AI can query:

  • Academic databases (PubMed, ArXiv, IEEE)
  • Internal document repositories
  • Web sources and news feeds
  • Personal note-taking systems

Example query: "Compare European AI safety research methodologies from 2023-2025 with US approaches" - AI retrieves and analyzes from multiple repositories.

Web Development and Content Management ​

Wix Platform Integration ​

Implementation: Wix adopted MCP to empower their AI web design tools through an embedded MCP server.

Capabilities:

  • Fetch site analytics and performance data
  • Modify page content and design elements
  • Create new website components
  • Integrate with marketing and e-commerce systems

Vision: Suggests a future where complex software (WordPress, Photoshop, etc.) might expose MCP interfaces for AI assistance.

DevOps and Cloud Management ​

AWS Official MCP Servers: Amazon released official MCP servers for:

  • AWS Lambda for serverless function management
  • Amazon ECS/EKS for container orchestration
  • Fargate for serverless container deployment

Microsoft Azure: Azure MCP Server connects AI agents with Azure services, enabling AI agents to act as junior cloud engineers.

Google Cloud: Launched fully-managed remote MCP servers (December 2025) for Maps, BigQuery, Compute Engine, and Kubernetes Engine—developers can now point AI agents to globally-consistent, enterprise-ready endpoints.

Monitoring and CI/CD: Integration with system health monitoring, alerting tools, and automated testing/deployment pipelines.

Docker MCP Catalog and Toolkit ​

Containerized MCP Ecosystem ​

Announcement: Docker embraced MCP with the Docker MCP Catalog and Toolkit, bringing container-grade security to AI agent tools.

Docker MCP Catalog: Now part of Docker Hub, featuring 100+ verified tools from partners including:

  • Cloud & APIs: Stripe, Elastic, Heroku
  • Data & DevOps: Neo4j, Pulumi, Grafana Labs
  • Monitoring: Kong Inc., New Relic
  • Development: Continue.dev, and many more

Key Benefits:

  • Each MCP server runs as an isolated, self-contained container
  • Portable, consistent execution without dependency issues
  • All servers are built and digitally signed by Docker
  • Works offline once downloaded

Docker MCP Toolkit ​

Integration: Management interface in Docker Desktop for setting up, managing, and running containerized MCP servers.

Dynamic MCP: Enables AI agents to discover, add, and compose MCP servers on-demand during conversations—no manual configuration required.

Client Support: Connect to Claude, Cursor, VS Code, Windsurf, Continue.dev, Goose, and the Docker AI Agent.

Governance and Industry Standardization ​

Agentic AI Foundation (AAIF) ​

December 2025: Anthropic donated MCP to the Agentic AI Foundation, a directed fund under the Linux Foundation.

Co-founders: Anthropic, Block, and OpenAI

Supporting members: Google, Microsoft, AWS, Cloudflare, Bloomberg

Significance: This move transformed MCP from a single-vendor specification into true industry infrastructure with neutral governance.

November 2025 Specification Updates ​

The November 2025 spec release introduced key enterprise features:

Extensions System: Components and conventions that operate outside the core specification, enabling scenario-specific additions without requiring full protocol integration.

OAuth 2.1 Authorization: Secure access to servers using standard authentication flows—critical for enterprise adoption with sensitive data.

Enterprise Registries: Vision for organizations to adopt self-managed MCP registries with governance controls and security coverage.

Community-Driven Innovation ​

MCP Server Registry Growth ​

Open-source ecosystem: The MCP ecosystem has grown to 5,800+ servers and 300+ clients. Popular servers include:

Popular services: Google Drive, Slack, GitHub, PostgreSQL databases, Stripe APIs

Development tools: Git repositories, Docker, web browsers (Puppeteer)

Specialized applications:

  • Cybersecurity tools and network monitors
  • Fun implementations like Tic-Tac-Toe game servers
  • Industry-specific solutions for logistics, manufacturing, retail

Example: Tic-Tac-Toe Tutorial ​

Educational project: GitHub published a tutorial "Building your first MCP server" demonstrating a turn-based game server.

Implementation: Server defines game-related tools (create game, make move, check state) and resources (list ongoing games).

Result: By connecting to VS Code, Copilot could play Tic-Tac-Toe by calling MCP tools, illustrating end-to-end capability extension.

Enterprise Deployment Patterns ​

Microservices Architecture ​

Pattern: MCP servers deployed as microservices within enterprise infrastructure.

Example setup:

  • Database MCP server (Docker container)
  • CRM system integration (authenticated API endpoint)
  • Email service connector (internal service account)

Security: Servers integrate with existing identity systems, using OAuth tokens or service accounts over HTTPS.

Multi-Cloud and Hybrid Deployments ​

Flexibility: Same MCP protocol works across:

  • Local servers via STDIO for on-device data
  • Remote servers via HTTP for cloud services
  • Hybrid architectures mixing both approaches

Scalability: Servers can sit behind standard web infrastructure (proxies, load balancers) and benefit from HTTP features like TLS encryption and status codes.

Development Time Savings ​

Integration Efficiency ​

Problem solved: Before MCP, developers wrote custom adapters for each combination of AI platform and external service.

MCP advantage: Build one MCP connector and gain compatibility with multiple AI assistants automatically.

Reported impact: Early success stories indicate MCP "drastically reduced the time to integrate AI with new data sources."

Cross-Platform Reusability ​

Example: A Google Drive MCP server built for Claude Desktop automatically works with:

  • OpenAI's ChatGPT Desktop
  • Custom applications using MCP SDKs
  • VS Code extensions and IDE integrations
  • Any future MCP-compatible AI platform

This reusability represents a fundamental shift from platform-specific plugins to universal connectivity.

Looking Ahead: 2026 and Beyond ​

Agent-to-Agent Communication ​

Current state: The MCP specification focuses on "Host-to-Server" communication—a single AI assistant connecting to multiple tools.

Next frontier: "Agent-to-Agent" communication, where multiple AI agents coordinate across dozens of MCP servers for complex multi-agent workflows.

Enterprise Scale ​

Current adoption: Enterprise teams are deploying MCP at scale with sophisticated security and governance controls. Some estimates suggest 90% of organizations will use MCP by the end of 2025.

Evolution: If 2025 was the year of adoption, 2026 will be the year of expansion. MCP is evolving into the standard infrastructure for contextual AI—the foundational layer that connects AI capabilities to real-world systems.

Sources ​

Interested in building your own MCP integration? Check out learning resources for official documentation, tutorials, and community support.

Community resource for Model Context Protocol knowledge and adoption.

© 2025 Model Context Protocol Belgium. Open information for AI advancement.