OtoDock Launches Self-Hosted 'Company OS' for Autonomous AI Agent Teams
The new platform allows organizations to deploy Claude Code and Codex agents within a sandboxed, hierarchical infrastructure on their own servers.
Dimitris Mourtzis has released OtoDock, a self-hosted "company OS" designed to let organizations deploy and manage teams of AI agents on their own infrastructure. The platform transforms high-capability CLI tools into collaborative digital employees that operate within a structured organizational hierarchy.
OtoDock integrates Anthropic's Claude Code and OpenAI's Codex, running them as persistent processes. To ensure security, the platform utilizes kernel sandboxes via bubblewrap and implements network isolation through pasta. This architecture allows agents to be organized into specific departments that mirror a company's actual organizational chart. Management is handled through a centralized dashboard that supports real-time monitoring, scheduled automation, and multi-agent meetings.
The Shift to Agentic Infrastructure
This release comes as the industry moves beyond simple chatbots toward autonomous "agentic" workers. While many AI tools operate as isolated sessions, there is a growing demand for infrastructure that provides persistence and collaboration. OtoDock addresses this by adding a management layer over existing CLI agents, allowing them to function as a cohesive workforce while utilizing the organization's existing AI subscriptions.
Enterprise Privacy and Control
By focusing on self-hosting and strict sandboxing, OtoDock targets enterprises and privacy-conscious businesses. The primary advantage is the ability to leverage autonomous coding and research agents without the need to send sensitive internal credentials or proprietary data to a third-party agent platform. This shift toward an "Agentic OS" architecture suggests a future where AI is not merely a tool used by a human, but a structured, integrated part of the corporate workforce.
Deployment and Access
OtoDock is available for self-hosting, with a free tier for teams of up to five users. Organizations requiring larger deployments must move to a seat-based licensing model. As the platform evolves, the industry will be watching how these sandboxed, departmentalized AI structures impact traditional software development lifecycles and internal corporate workflows.