Ink, also known as ml.ink, is a cloud infrastructure platform designed for use by artificial intelligence (AI) agents. Launched on March 5, 2026, the service enables AI agents to autonomously deploy, manage, and scale the entire lifecycle of web applications and services. ^[5] The platform's primary innovation is the Model Context Protocol (MCP), an application programming interface (API) that exposes all infrastructure capabilities as callable tools, allowing AI models to programmatically manage resources without human intervention through traditional dashboards or continuous integration/continuous delivery (CI/CD) pipelines. ^[5]

Overview

Ink functions as an infrastructure-as-a-service (IaaS) provider built specifically for an agent-centric workflow. ^[4] The platform is designed to abstract complex DevOps operations, enabling deployments through natural language prompts given to an integrated AI agent. Its motto is "Built for humans. Designed for agents." ^[3] The core operational model allows a user to connect their Ink account to a Model Context Protocol (MCP) compatible client using a generated API key. From there, the user can instruct the agent in plain language to build a specific application or infrastructure. The agent then autonomously utilizes Ink's MCP tools to execute tasks such as creating code repositories, provisioning databases, deploying services, and configuring domains. ^[3]

Ecosystem and Integrations

Ink is designed to integrate with a growing ecosystem of AI tools and services.

Supported MCP Clients

The platform is compatible with any AI client that supports the Model Context Protocol. As of its launch, supported clients included:

• Claude
• Cursor
• VS Code
• Codex
• Gemini CLI
• Windsurf
• Goose
• Kimi Code
• Open Code
• Cline ^[3]

Use Cases

The platform's architecture is suited for a wide range of web application and service deployments, including:

• SaaS Applications: Full-stack applications with frontend, backend, and database components.
• API Services: Deployment of REST or GraphQL APIs.
• AI Agent Backends: Hosting for services that power AI agents, including tool servers and model inference endpoints.
• Landing Pages: Static sites and Single Page Applications (SPAs).
• Internal Dashboards: Data visualization and administrative panels built with tools like Streamlit and Gradio.
• MCP Tool Servers: Hosting for custom-built MCP tools that extend the capabilities of AI agents. ^[3]