MintMCP helps companies put agents to work safely. We exist to help companies do more of their work with agents, and we build the infrastructure that makes a company’s own work runnable by them. That has to start from trust. Our first product is an MCP gateway that gives those agents governed access to company data and tools.
Before agents can touch Slack, GitHub, Jira, Snowflake, internal databases, or production systems, you need to know what they can reach, what they did, and why they were allowed to do it. Most people carry far more permissions than any single task needs. If an agent uses your token directly, it inherits all of them. MintMCP scopes that access down to exactly what an agent needs to do its job and captures the data flows and audit logs security teams need. From there we are building the rest of the agent stack: connectors, monitoring, sandboxing, long-running agents, and memory, with the goal of becoming the system a company’s work runs on.
We work in person in the San Francisco Bay Area. The structure is flat and the market moves fast, so you have a lot of latitude over what you work on and how.
We are AI-native. Everyone uses Claude and Codex every day, and we have more agents than employees. They run in Slack alongside the team and handle sales, marketing, deal ops, and engineering on the same gateway we sell to customers, and we are building an entire company operating system around them.
Since launching in February 2026, we have grown to 50+ paying customers, with strong inbound demand. Customers include Coursera, Arlo, and Braze. We are backed by Cowboy Ventures and Coatue, with angels including Frederic Kerrest (Okta), Michael Grinich (WorkOS), Andrej Karpathy, Jeff Dean, and Scott Belsky. Our founders are machine learning experts from Stanford, Berkeley, and CMU, with applied research at Google Brain and Waymo, and one co-founded Coursera.