Show HN: Self-hosted company OS, Claude Code and Codex agents in departments
summary
OtoDock is a self-hosted platform designed to act as a central brain for a company by managing collaborative AI agents. It allows businesses to create specialized agents with distinct personas and tools that can work independently, delegate tasks to one another, and integrate with existing company hardware and software. By utilizing the user's own API subscriptions and local models, it ensures data control and flexibility in AI engine selection.
key points
- The platform features a highly customizable agent architecture consisting of persona, memory, memory, workspace, knowledge, skills, and tools, supporting four distinct sharing modes for team collaboration.
- Security is prioritized through the use of kernel sandboxes and network isolation for every agent session, ensuring that AI agents cannot access unauthorized internal services or sensitive credentials.
- OtoDock extends AI capabilities beyond text by integrating telephony for phone calls, remote machine pairing for local hardware control, and a community catalog of ready-made tools and agents.
who should read it
Business owners and technical managers who want to deploy secure, self-hosted AI agent workflows to automate company operations without relying on a third-party cloud OS.
discussion positive
consensus
Commenters generally agree that autonomous agent systems should prioritize deterministic backbones and specialized tools over relying solely on LLM 'magic'. There is a consensus that a successful implementation must avoid becoming a 'slop factory' by using strongly typed code and existing integrations via MCP rather than reinventing basic functionality. The sentiment is cautiously optimistic, provided the architecture remains tool-driven and decentralized.
top insight
The most valuable perspective is that an idealized autonomous company should be decentralized and tool-driven rather than agent-driven. By pushing LLM capabilities to the edge of specific tools and maintaining a deterministic core, developers can ensure consistency and prevent the system from losing coherence through excessive token consumption and memory mismanagement.
Generated by an LLM from the article and thread. Check the sources before relying on it.