Mintive launched on Thursday as the infrastructure for what it calls the agentic enterprise, announcing $11 million in total funding including a new $7 million seed round led by Cowboy Ventures. The startup gives each AI agent its own identity, only the access its work requires, and a record of everything it does. It says its platform already processes 2.5 million agent calls a day across more than 50 paying enterprise customers.
What Mintive sells
Mintive’s platform has five products as one control layer for enterprise agents. An MCP gateway governs which tools an agent may reach, an LLM gateway routes model calls and sets budgets, and an agent identity layer gives each agent a scoped, revocable credential delegated from the person it works for. An agent monitor records the chain behind every action.
The fifth product, coworker agents, runs inside Slack and Teams, but only in rooms it has been permitted to enter. A screening layer called Mint Guard checks calls for prompt injection, exposed secrets and personal data before they reach company systems. Every call is logged, and the platform is SOC 2 Type II audited, with EU and US data residency.
The idea is to treat an agent like an employee: it gets an identity, permissions and an audit trail, whatever model sits underneath, the same identity-first approach as IBM’s verified agent identities. That framing matters because agents act across systems on a user’s behalf, and an over-privileged agent can read, change or send far more than its task requires.
Who is behind it
Mintive was founded by chief executive Jiquan Ngiam, Coursera’s first employee and later a Google Brain researcher, and chief technology officer Vijay Vasudevan, one of the creators of TensorFlow at Google Brain. The funding round drew Coatue, Operator Collective and Hustle Fund, alongside angels including Okta co-founder Frederic Kerrest, former Google chief scientist Jeff Dean and OpenAI co-founder Andrej Karpathy.
Cowboy Ventures managing partner Aileen Lee said she was drawn to the founders’ restraint on round size, contrasting Mintive’s $11 million total with agentic-infrastructure peers raising far larger sums at unicorn valuations. The company launched its product in February and rebranded from MintMCP to Mintive in October, as it widened its product set from a gateway into an agent platform.
Customers and traction
Mintive names Ro, Modern Treasury, Braze, Coursera, the University of Michigan, Apax and Stability AI among its more than 50 customers, and says it has handled more than 200 million agent calls in seven months. At Modern Treasury, an agent searches tools before an account manager steps in, saving between 30 minutes and four hours per case, per the company.
At the University of Michigan, Mintive is the MCP gateway across 150,000 community members, giving IT one path to approved tools and one place to control access. At Ro, it operates in a healthcare environment where agent access and auditability carry compliance requirements alongside the work. Those figures are company-reported, and Mintive has not published verification of its call volumes.
The customer list is the strongest part of the announcement, because enterprise agent governance is easy to pitch and hard to deploy. Winning regulated and large-organisation customers in eight months suggests the product solves a real procurement problem: security teams will approve agents only when they can see, limit and audit what those agents do.
Why It Matters
Agent identity is becoming a distinct infrastructure category, sitting between the model providers and the enterprise systems agents touch. Amazon, Microsoft, OpenAI and Anthropic all offer agent-management features, and point products crowd the same space, so Mintive is entering a fiercely competitive market. Its bet is that a model-agnostic control layer wins over tools tied to one provider’s models.
The same governance question is driving vendors too, as shown by Akamai’s safe execution environment for agents. Mintive’s smaller, focused pitch is speed of deployment: the company says customers are surprised by how quickly they can roll it out, because access control, permissioning and memory management are handled underneath rather than configured by hand.
Investors are willing to fund that pitch, but the $7 million round is small by current AI standards, and Mintive will need to prove it can scale beyond its first 50 customers without being absorbed into a platform vendor’s bundle. The next test is whether its governance layer becomes a standard purchase or a feature that larger vendors give away.
Background
Analyst firm Gartner expects the average Global Fortune 500 company to run more than 150,000 AI agents by 2028, up from fewer than 15 in 2025, according to Mintive’s launch post. If that forecast is even broadly right, the question of who controls those agents becomes a central enterprise security problem, not a niche concern for early adopters.
Source: Mintive announcement, TechCrunch report
0 Comments