Enterprise MCP gateway that brokers AI assistant access to internal tools and data

MintMCP, which operates as Dependable AI Inc. out of San Francisco, was founded by Jiquan Ngiam, Vijay Vasudevan, and Bozhi See, and launched its enterprise governance platform publicly in February 2026. The problem it targets is familiar to anyone who has watched Model Context Protocol adoption inside a large company: individual engineers run MCP servers locally, paste long-lived API keys into configuration files, and connect assistants such as Claude, Cursor, and ChatGPT to production systems with no record of what was accessed. MintMCP replaces that pattern with a gateway that security and platform teams operate.

Local or containerized MCP servers become hosted services behind OAuth and single sign-on, with credentials held by the gateway and access granted per user, team, or role. A catalog covers common enterprise connections including Gmail, Outlook, Salesforce, and Snowflake, with a much larger registry of community servers available for review before approval. The agent monitor gives real-time visibility into which tools an assistant called and with what arguments, applies guardrails such as secret scanning and personal data detection, and produces logs that can be exported for audit. The company holds SOC 2 Type II certification, lists Coursera, Braze, and Modern Treasury among its customers, and counts Coatue, Hustle Fund, Andrej Karpathy, and Jeff Dean among its backers.