Opening Internal Operations to AI: Air Closet Unveils Suite of 17 MCP Servers
infrastructure#agent📝 Blog|Analyzed: Apr 8, 2026 05:32•
Published: Apr 7, 2026 23:02
•1 min read
•Zenn ClaudeAnalysis
This is a fascinating real-world case study of enterprise AI integration, showcasing how Model Context Protocol servers can bridge the gap between large language models and complex internal business tools. The sheer scope of the implementation—covering everything from databases and infrastructure to project management and CI/CD—demonstrates a bold commitment to operational efficiency. By enabling non-engineers to deploy code and interact with data safely, they are pioneering a truly democratized development environment.
Key Takeaways
- •The company deployed 17 distinct MCP servers within 3 months to connect AI with internal tools like AWS, GCP, and Google Workspace.
- •The architecture prioritizes security and modularity by separating servers based on authentication scopes (e.g., isolating Workspace API permissions from database queries).
- •The system enables 'Sandbox' environments where non-engineers can use AI to modify code and deploy applications safely.
Reference / Citation
View Original"We have built 17 MCP servers, including DB Graph, currently in operation internally. Spanning DB, infrastructure, documents, project management, Observability, CI/CD, and even code modification and deployment by non-engineers, we have reached a state where AI can operate across a very wide range of internal business domains."
Related Analysis
infrastructure
Maximizing 8GB VRAM: Why Multi-Model Local LLM Setups Outperform Single Giants
Apr 7, 2026 23:00
infrastructureImplementing the AI Improvement Loop: A Blueprint for Review Infrastructure and Root Cause Analysis
Apr 8, 2026 00:31
infrastructureSpec-Driven Development: Designing SaaS as Interchangeable Components
Apr 7, 2026 22:45