THE ESSENTIALS IN 30 SECONDS
- Cowboy Ventures led the $7 million seed round, bringing Mintive's total funding to $11 million; Coatue, Operator Collective and Hustle Fund also invested.
- Angels include Okta co-founder Frederic Kerrest, former Google chief scientist Jeff Dean and OpenAI co-founder Andrej Karpathy.
- The platform, launched in February, combines an MCP gateway, an LLM gateway with cost caps, separate agent identities and company-wide agent monitoring; Mintive reports more than 50 enterprise customers.
Mintive, a startup building tools for companies to control the AI agents their employees deploy, has raised a $7 million seed round led by Cowboy Ventures, TechCrunch reported on October 29. The new money brings its total funding to $11 million.
Who is behind it
Coatue, Operator Collective and Hustle Fund joined the round. The list of individual backers stands out: Okta co-founder Frederic Kerrest, Jeff Dean, Google's former chief scientist, and Andrej Karpathy, a co-founder of OpenAI.
The founders also come from Google Brain. Chief executive Jiquan Ngiam was Coursera's first engineer before working on neural networks at Google Brain, and chief technology officer Vijay Vasudevan co-created the TensorFlow machine learning framework there.
What the product does
Mintive sells an agent management platform aimed at IT departments. Its main components, as described by TechCrunch, are:
- an MCP gateway that decides which cloud services agents may reach and which actions they may take, using the Model Context Protocol that many agents rely on to connect to tools;
- an LLM gateway that routes requests to different models and can cap spending;
- identities and permissions for agents that are separate from those of the employees they work for;
- a monitor that tracks every agent across the company, along with tools for building agents that carry out long-running tasks on their own.
The company says setup takes a few clicks. The product launched in February 2026, and Mintive reports more than 50 enterprise customers, including Ro, Modern Treasury, the University of Michigan, Coursera, Apax and Stability AI. These figures have not been independently confirmed.
A crowded field
Giving agents their own identity, rather than letting them act with a human employee's full access, is becoming a common security recommendation as companies connect agents to email, code repositories and business software. Mintive faces competition from Amazon, Microsoft, OpenAI, Anthropic and many startups offering similar controls, TechCrunch notes. Its seed-stage funding is modest next to those players, so the founders' track record and early customer base will be tested as larger platforms bundle agent governance into their own products.
