MintMCP, now renamed Mintive, builds the control layer between a company’s AI agents and the tools, data and models they use. It is for security and platform teams that want employees and autonomous agents to use MCP servers and large language models without handing agents human passwords or losing track of what they did.
The problem it addresses is agent sprawl: tools adopted team by team, each with its own credentials and no shared record of what was done.
At A Glance
| Vendor | Mintive (formerly MintMCP) |
|---|---|
| Category | MCP and LLM gateway with identity and monitoring for AI agents |
| Identities covered | Human users via SSO and SCIM, autonomous AI agents, downstream service credentials |
| Deployment | Managed cloud service with EU and US data residency; on-premises on request |
| Customers | Coursera, Stability AI, Modern Treasury, Eurostar, Braze |
| Certifications | SOC 2 Type II, HIPAA |
| Market classification | Agentic AI Identity; also AI Governance & Compliance |
What MintMCP Does
When agents borrow people’s credentials and connect straight to company systems, nobody controls which tools they reach or can trace an action back to the agent that took it. Mintive addresses this by routing all agent and user traffic through one governed layer: administrators decide which MCP servers and models are allowed and for which roles, each agent acts under its own identity, and every call is logged. The same controls apply whether a request comes from an employee’s assistant or a fully autonomous agent, so security teams manage one policy rather than one per tool.
Products
MCP Gateway
The MCP gateway connects AI tools such as Claude, ChatGPT, Cursor and Copilot to company data through a single endpoint. Administrators approve servers, bundle them by role and switch off destructive tool actions, so a sales team’s assistant can read the CRM without being able to delete records. Hosted connectors run MCP servers in Mintive’s own infrastructure, which removes local installs on laptops, and a private tunnel reaches servers inside the customer network.
Agent Identities
Agent identities stop agents from reusing human logins. Each agent receives scoped, revocable credentials, presented as a bearer key or M2M token and exchanged for short-lived access tokens. Every call is attributed to the agent and records the administrator who issued it, which gives auditors a clear owner for each automated action. Deleting a bearer key blocks the agent’s next request immediately, deleting an agent removes all its credentials and downstream authorisations in one step, and agents cannot call the Admin MCP, change rules or manage other agents, so an agent cannot widen its own access.
Agent Monitor
The agent monitor shows security teams what their agents actually did. Tool calls pass through Mint Guard, which screens them for prompt injection, secrets and personal data before they reach downstream systems.
LLM Gateway And Coworker Agents
The LLM gateway sets which models each team may use and how much it may spend, tracking spend and tokens so finance and platform owners can control model costs. Coworker agents are AI teammates that work in Slack and Teams, so staff can hand work to an agent in the chat tools they already use, under the same access rules as the gateway.
How It Works
Mintive runs as a managed cloud service, with on-premises or self-hosted set-ups available on request.
| Component | What it does |
|---|---|
| Virtual MCPs | Logical groupings of connectors are exposed as single endpoints, so a client connects to one endpoint instead of many servers. |
| Connectors | Hosted, remote or STDIO servers talk to downstream systems. Hosted connectors run on Fly.io Machines in Mintive’s managed data plane with no inbound route from the internet, and open-source STDIO and custom packaged servers run as containers. |
| Request flow | A client authenticates, Mintive checks identity and policy before returning the permitted tools, then routes each tool call to the right connector and logs it. |
| Credentials | Connectors never hold long-lived credentials. Mintive injects them per call and encrypts stored credential material at rest with rotated AES keys. |
| Audit | Audit logs, access policies and credential management sit in one place and can be exported to a SIEM. |
Identities And Systems Covered
| Human users | Sign in with SSO and are provisioned through SCIM; servers work as soon as SSO completes. |
|---|---|
| AI agents | Machine-to-machine identities that see only the MCPs created for them and cannot log in through SSO or complete interactive OAuth. |
| Downstream credentials | Shared by default, with agent-owned credentials where a connector needs them. |
| Systems | Gmail, Outlook, Google Calendar, Outlook Calendar, Linear, Notion, Snowflake, Elasticsearch, Salesforce and databases. |
Company Background
| Co-founders | Include Jiquan Ngiam and Bozhi See. |
|---|---|
| Funding | $11 million in total, including a round led by Cowboy Ventures with Operator Collective, Burst Capital, Coatue, Hustle Fund and Maven Ventures. |
| Scale | More than 50 reported customers and 2.5 million agent calls a day. |
Note: This overview is compiled from MintMCP’s publicly available information as of 3 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.