Palma

Palma provides an enterprise‑level governance and cost‑control layer for AI agents that interact with core business systems through the Managed Call Protocol (MCP). It sits between agents and the underlying services, pruning and scoping context before execution, attributing usage to specific agents or departments, and enforcing policies, guardrails, and human‑in‑the‑loop approvals at the tool boundary. The platform records every tool call, enabling auditability, cost tracking, anomaly alerts, and compliance reporting.

The system is intended for AI development teams, MCP administrators, and organizations that deploy autonomous or chat‑based agents across a range of tools such as databases, issue trackers, design platforms, and cloud services. By registering internal or vendor MCP servers with Palma, teams can manage per‑tool policies, approvals, and ROI metrics without modifying each integration individually.

Palma is an experimental, open‑standard solution that can be deployed on‑premises or within a VPC, supporting existing enterprise LLMs and allowing agents to operate autonomously while keeping data usage, security, and performance under centralized control.