Faramesh – The first deterministic execution control plane for AI agents (github.com)

Faramesh has introduced its Core, Horizon, and Nexus offerings, marking the launch of the first deterministic execution control plane for AI agents. Faramesh Core features a policy-driven governance system that includes risk scoring and human-in-the-loop approval for AI agent actions. Horizon is a fully-managed SaaS solution designed for rapid onboarding without the need for infrastructure management, while Nexus offers an enterprise-grade deployment that enhances security and management controls, such as SSO and audit capabilities. This suite emphasizes a "deny-by-default" security model with YAML-based policies that allow organizations to set flexible governance rules. The significance of Faramesh lies in its ability to provide meaningful oversight for AI agents, balancing autonomy with security. Its innovative risk scoring system automatically assesses the risk associated with each action, potentially requiring approvals even for actions that would normally be allowed. With SDK integrations for Python and Node.js, as well as comprehensive auditing features like an event timeline and detailed action summaries, Faramesh aims to streamline the governance of AI tools, making it easier for developers and organizations to maintain control while benefiting from AI's execution capabilities.