Bitgo Holdings Inc. (NYSE: BTGO) announced the launch of its Model Context Protocol (MCP) Server on Monday, a new gateway designed to connect the company's institutional cryptocurrency infrastructure directly with AI-powered development environments. The tool allows developers and AI agents to interact with Bitgo's technical documentation using natural language, eliminating the need to manually browse through pages of API references.
How the MCP Server Works
The MCP Server integrates with popular AI platforms including ChatGPT, Claude, Cursor, JetBrains IDEs, Visual Studio Code, and Windsurf. After adding a single endpoint to an MCP-compatible client, developers can ask questions like “How do I create a wallet?” or “Which endpoints support staking?” The system retrieves answers directly from Bitgo's official documentation, generates code snippets, and explains API functionality on demand.
Bitgo emphasized that the server is currently limited to documentation retrieval and contextual support. It does not execute transactions, manage custody, or perform on-chain operations through AI agents. This restriction is deliberate, according to the company, reflecting a careful first step toward autonomous crypto agents. While the broader industry has flirted with fully self-directed crypto AI agents, Bitgo keeps its initial iteration firmly in the “read and assist” lane rather than “act and transact.”
CEO Mike Belshe framed the launch as a foundational move. “AI is changing how developers build, navigate technical systems, and interact with infrastructure,” he said. “Developers can now treat Bitgo as agent infrastructure—and this is just the first step toward making our platform fully accessible to the AI economy. Much more is to come.”
Security-First Design for Institutional Use
The MCP Server uses a public endpoint linked to Bitgo's live documentation system, ensuring that AI get access to up-to-date materials rather than relying on static training data. The company also noted that its Developer Portal already includes an embedded “Ask AI” feature on documentation pages, allowing users to query content without setting up external tools.
Bitgo's approach is tailored to the institutional sector, focusing on custody, wallets, staking, and settlement. If AI agents are to manage large capital flows, they will need infrastructure built for compliance and scalability. The MCP server handles the basics—faster development, cleaner integrations, and fewer excuses for developers to get lost in documentation.
Broader Industry Context
The MCP standard, introduced by Anthropic, is gaining traction in both AI and crypto circles. It provides a common framework for connecting AI systems to external data sources, reducing the need for custom integrations. Bitgo's implementation is one of the first in the crypto space to offer a production-ready MCP endpoint for institutional developers.
The launch comes amid a broader industry shift where platforms are racing to make their services AI-native. Competitors and adjacent projects are experimenting with similar MCP-based systems, particularly for trading and on-chain interactions. However, Bitgo's focus on institutional compliance and security may give it an edge in the race to deploy AI agents for high-value crypto operations.
For now, the MCP Server is available immediately to existing Bitgo developers at no additional cost. Setup instructions and examples are published in Bitgo's developer documentation. Belshe's hint at “more to come” suggests that subsequent iterations may expand the server's capabilities beyond read-only access, potentially allowing AI agents to execute limited actions under strict guardrails.