Putting agents to work takes more than connecting tools. Teams need clear control over what agents can reach, a way to stop unsafe actions before they run, a record of what every agent did and what it cost, and agents that have long-term memory and do real work.

MintMCP is now Mintive.

We launched MintMCP early in 2026 as an MCP gateway. Since then, we've shipped Mint Guard, Agent Monitoring, an LLM gateway, and Coworker Agents. Mintive brings that full product range under one name: MintMCP is now Mintive.

MCP Gateway: Governed Access for Every AI Tool

Approve the servers your team can use, and block the rest. By grouping servers into role-based bundles with destructive actions turned off, you can have fine-grained control on what tools are available so that agents reach only what you permit.

Those controls are already running at customer scale: Coursera uses them to govern AI for the entire company, Deerfield Group used them to replace scattered, duplicate tools with one vetted marketplace for the whole agency; Deerfield's CTO called it the fastest rollout he'd seen.

LLM Gateway: Cost and Usage Control for Model Calls

The MCP gateway governs what agents can reach. The LLM gateway covers the other half of their traffic: the model calls themselves. Route Claude Code, Codex, Claude Desktop, and Coworker Agents through it, and every call appears with its usage, cost, model, and latency next to the agent's tool calls.

Give each team its own budget and list of allowed models, using the groups you already manage in Mintive. When a team hits its budget, new calls stop. Connect OpenAI, Anthropic, or Fireworks and switch between them without rewiring each agent.

Mint Guard: Real-Time Policy for Every Tool Call

Mint Guard checks every tool call in real time for prompt injection and unsafe operations. We maintain the detection policy, so your team does not have to write rules. Start in monitoring mode to see what would be blocked, then switch to enforcement. For more advanced guardrails, use middleware to introduce custom scanning logic both before and after tool calls.

Agent Monitoring: A Record of Agent Work

Every call traces back to a person, agent, server, and model. See who's using agents, track spend by user, model, and tool, and review any session call by call with timestamps, arguments, and results. Export to your SIEM, or stream over OTLP.

Coworker Agents: AI Teammates in Slack

Coworker Agents is our newest and fastest-growing product. Each agent works like a teammate: it lives in the Slack channels your team already uses, remembers what it learns, and gets better at the job over time.

• Long-term memory. Coworkers carry what they've learned from one run to the next, so they can pick up where they left off without asking the team to re-explain the job. That memory lives in files your team can read and edit.
• Dynamic permissions. What a coworker can access depends on the channel it's in and who's messaging it, so the same agent can do more for the team that owns a system and less everywhere else.
• Proactive work. Coworkers wake up on their own: they run on a schedule, and they jump back into threads they're part of when there's something useful to add.

Each coworker has its own identity, and its tool calls run through the same gateway and Agent Monitoring as the rest of your agents. Read more about how Workstream, Modern Treasury, and Bared Footwear are deploying agents today.

Built for the Agentic Future Ahead

We are excited to introduce our broader product lines for teams putting enterprise agents into production. Everything customers run today carries over to Mintive, with more launches coming soon.

Interested in any of these products? Get a demo and we'd love to walk you through them.