Awake Venture Studio has announced the launch of ForgeOS, an open-source multiplatform operating system designed specifically for AI agents. Developed within Making Science, ForgeOS is positioned as the first agentic harness focused on governance, permissions, budgeting, and oversight for enterprise AI systems.
The platform is designed to help organizations manage autonomous AI agents operating across major AI ecosystems including Google, Anthropic, OpenAI, and other frameworks without requiring changes to existing agent code.
As organizations increasingly deploy autonomous AI agents across business operations, governance and oversight have become major operational concerns.
According to Awake Venture Studio, existing AI platforms provide tools for building agents but lack mechanisms for controlling spending, permissions, compliance, and operational accountability.
ForgeOS was developed to address these challenges by acting as a governance layer between AI agents and the underlying AI models they use.
“With a kernel architecture modeled on UNIX, ForgeOS is the first agentic harness — a governance layer that enforces budgets, permissions, and human oversight for AI agents on any framework,” the company stated.
The platform is designed to help enterprises manage AI agents operating across departments including sales, marketing, finance, operations, and infrastructure management.
ForgeOS functions similarly to a traditional operating system by controlling how AI agents interact with systems, tools, and enterprise data.
The platform provides centralized governance controls regardless of which AI provider powers the underlying agent.
Key capabilities include:
Organizations can define daily or task-level spending limits for individual AI agents. Once limits are reached, agents automatically stop execution.
ForgeOS enables enterprises to restrict what actions agents can perform and what systems or data they can access.
For example:
The platform allows enterprises to require human approval for sensitive or regulated actions before execution.
Examples include:
ForgeOS records every action, decision, and approval made by AI agents to support compliance, transparency, and regulatory oversight.
If an AI agent fails during execution, the system allows workflows to resume from the interruption point rather than restarting tasks entirely.
According to the company, ForgeOS supports governance across multiple AI ecosystems simultaneously, including:
The company stated that organizations increasingly rely on multiple AI vendors for different workloads and need consistent governance policies across all environments.
“ForgeOS applies the same governance rules to all of them through a single control panel,” the company said.
ForgeOS is already being used internally by Making Science across production AI workloads and operational systems.
The company said one of the first deployments involved an AI agent auditing cloud infrastructure across hundreds of projects in minutes.
Additional production deployments include:
The company stated that all four production agents are also available as open source.
“We built ForgeOS because we needed it ourselves,” said José Antonio Martínez Aguilar, CEO and Founder of Making Science. “When you run AI agents for clients, you need to guarantee that Agent A cannot access Client B's data, that spending stays within budget, and that sensitive decisions get human review. No AI platform offered this. So we built it.”
ForgeOS has been released as open source to support transparency, auditability, and enterprise trust.
The company stated that the platform aligns with emerging AI governance requirements, including oversight and accountability provisions outlined in the EU AI Act.
“We believe AI governance should be transparent,” Martínez Aguilar added. “Companies trusting AI agents with their operations and their clients' data deserve to see exactly how those agents are controlled. Open source is not just a licensing choice — it is a trust signal.”
ForgeOS also includes Mission Control, a real-time dashboard that provides visibility into active agents, operational spending, task execution, and approval workflows.
ForgeOS is available immediately on GitHub.
The full technical vision for the platform is also available through “The ForgeOS Manifesto” published on GitHub.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones, and thrive on the tech pulse.
