Run company departments with persistent self-hosted agents
OtoDock · Collaborative Agents. Task: OtoDock · Collaborative Agents Your team builds them on Claude Code, Codex or local models, and they work in chats, on schedules, on the phone and on your own machines. The agentic company OS. Install · Video · Docs · Website · Features · Discussions · X · LinkedIn Dashboard highlights. Watch the two-minute video, made by an OtoDock agent. Why OtoDock OtoDock acts as the brain of your company. You create powerful agents that connect to the tools your company runs on, work in departments, delegate to each other, and keep working on their own when no one is watching. Your agents can build and deploy internal apps for your team, or even public ones you share with your customers. OtoDock is multi-tenant by design. Many people work with the same agents, and four modes decide how an agent is shared and where its work lands. You work with your agents on your self-hosted dashboard, or you can even give them a phone line and talk to them. All of it runs on your own Anthropic and OpenAI subscriptions, and on local models. OtoDock is built, tested and shipped by agents running on OtoDock. Quick start You need a Linux server (x86-64 or arm64) with Docker Engine 26 and Docker Compose 2.26 or later, or a Mac with Apple silicon and Docker Desktop, with at least 4 GB of RAM (sizing) and about 13 GB of free disk, plus a Claude or ChatGPT subscription, an API key or a local model server. Then, in its own folder: The script does fresh installs only, and scripts/upgrade.sh upgrades one later. The installation guide covers the rest: the server's address, HTTPS behind a reverse proxy, the phone service and building from source. Then open http://localhost:8400, or your server's address on port 8400, and the setup wizard greets you. Create the owner account, and OtoDock installs the Personal Assistant for you, with the tools it needs. Connect your Claude or ChatGPT subscription under Setup, AI Engines. Then open the Personal Assistant from the Agents page and send it your first message (first run). From there you add agents from the community catalog or build your own, and you add the people who will work with them. Build your own AI agents Every agent is a digital employee you give a job. It has its own knowledge, workspace, skills and tools. The Agents page in the docs explains each part. Four ways to share one agent Every agent has a mode that decides how people share it, and where its workspace lives: private for each person, shared by the whole team, or both. Everyone gets the right seat Each person has a role on every agent they use. The role decides what they can see and do there. Admins add each person to the agents they need. Claude Code or Codex on your subscription, or local models. Claude Code runs on your Anthropic subscription, and Codex runs on your ChatGPT subscription. Every user connects their own subscription or API key, or an admin shares theirs with the team. Local models on your own hardware work as well, through Ollama or any OpenAI-compatible server. You pick the engine per agent, and you can switch it per chat (AI engines). They work while you are away Agents run on any schedule, when a webhook event fires, or once at a time you choose. They work in your workspace, where their reports and files land. They notify you when something needs you, from a quiet chime to a persistent danger alarm (tasks, triggers). Every run is a full chat you can open, read and continue. The checks you attach judge the work at the end of a turn, and send the agent back for another round when it falls short (checks). Runs on your server. Works on your machines. The same agent, with the same files and tools, moves to any of your computers and keeps working there. By default, all your agents run on your server, in an isolated sandbox, with one dashboard controlling all of them. On your computer, the agent has full access to the machine, and you still work with it from the same dashboard, anywhere. You pair a laptop, a workstation or a PC running macOS, Linux or Windows with a one-line install. The machine keeps a single outbound connection to your server, so it needs no open ports, and its files stay in sync. If the machine goes offline, your server takes over (remote machines). Your own cloud of agents. Build and deploy full apps, just by asking your agent.; tools: Claude Code, Codex. Human role: The original post does not say how a person stayed involved.; limitations: Requires a Linux server or Apple silicon Mac with Docker Engine 26+, Compose 2.26+, 4GB RAM, and 13GB disk. Users must provide their own Anthropic/OpenAI subscriptions or API keys. Self-hosting is free for up to 5 users; larger teams require licensing. Agents run in isolated sandboxes with network restrictions. Human approval is necessary for critical actions, and the source does not specify automated failure recovery protocols beyond quality checks..
Tools
Published on Agent Work Atlas:
Overview
OtoDock · Collaborative Agents
Workflow
OtoDock · Collaborative Agents Your team builds them on Claude Code, Codex or local models, and they work in chats, on schedules, on the phone and on your own machines. The agentic company OS. Install · Video · Docs · Website · Features · Discussions · X · LinkedIn Dashboard highlights. Watch the two-minute video, made by an OtoDock agent. Why OtoDock OtoDock acts as the brain of your company. You create powerful agents that connect to the tools your company runs on, work in departments, delegate to each other, and keep working on their own when no one is watching. Your agents can build and deploy internal apps for your team, or even public ones you share with your customers. OtoDock is multi-tenant by design. Many people work with the same agents, and four modes decide how an agent is shared and where its work lands. You work with your agents on your self-hosted dashboard, or you can even give them a phone line and talk to them. All of it runs on your own Anthropic and OpenAI subscriptions, and on local models. OtoDock is built, tested and shipped by agents running on OtoDock. Quick start You need a Linux server (x86-64 or arm64) with Docker Engine 26 and Docker Compose 2.26 or later, or a Mac with Apple silicon and Docker Desktop, with at least 4 GB of RAM (sizing) and about 13 GB of free disk, plus a Claude or ChatGPT subscription, an API key or a local model server. Then, in its own folder: The script does fresh installs only, and scripts/upgrade.sh upgrades one later. The installation guide covers the rest: the server's address, HTTPS behind a reverse proxy, the phone service and building from source. Then open http://localhost:8400, or your server's address on port 8400, and the setup wizard greets you. Create the owner account, and OtoDock installs the Personal Assistant for you, with the tools it needs. Connect your Claude or ChatGPT subscription under Setup, AI Engines. Then open the Personal Assistant from the Agents page and send it your first message (first run). From there you add agents from the community catalog or build your own, and you add the people who will work with them. Build your own AI agents Every agent is a digital employee you give a job. It has its own knowledge, workspace, skills and tools. The Agents page in the docs explains each part. Four ways to share one agent Every agent has a mode that decides how people share it, and where its workspace lives: private for each person, shared by the whole team, or both. Everyone gets the right seat Each person has a role on every agent they use. The role decides what they can see and do there. Admins add each person to the agents they need. Claude Code or Codex on your subscription, or local models. Claude Code runs on your Anthropic subscription, and Codex runs on your ChatGPT subscription. Every user connects their own subscription or API key, or an admin shares theirs with the team. Local models on your own hardware work as well, through Ollama or any OpenAI-compatible server. You pick the engine per agent, and you can switch it per chat (AI engines). They work while you are away Agents run on any schedule, when a webhook event fires, or once at a time you choose. They work in your workspace, where their reports and files land. They notify you when something needs you, from a quiet chime to a persistent danger alarm (tasks, triggers). Every run is a full chat you can open, read and continue. The checks you attach judge the work at the end of a turn, and send the agent back for another round when it falls short (checks). Runs on your server. Works on your machines. The same agent, with the same files and tools, moves to any of your computers and keeps working there. By default, all your agents run on your server, in an isolated sandbox, with one dashboard controlling all of them. On your computer, the agent has full access to the machine, and you still work with it from the same dashboard, anywhere. You pair a laptop, a workstation or a PC running macOS, Linux or Windows with a one-line install. The machine keeps a single outbound connection to your server, so it needs no open ports, and its files stay in sync. If the machine goes offline, your server takes over (remote machines). Your own cloud of agents. Build and deploy full apps, just by asking your agent.
Reported result
No measured business outcomes or success metrics are reported in the source. The text describes the platform's capabilities and installation steps but does not provide evidence of specific operational improvements, cost savings, or task completion rates achieved by users. Acceptance should be verified by confirming successful agent deployment, connectivity to AI engines, and execution of a scheduled task with a passing quality check.
Human role
The original post does not say how a person stayed involved.
Limitations
Requires a Linux server or Apple silicon Mac with Docker Engine 26+, Compose 2.26+, 4GB RAM, and 13GB disk. Users must provide their own Anthropic/OpenAI subscriptions or API keys. Self-hosting is free for up to 5 users; larger teams require licensing. Agents run in isolated sandboxes with network restrictions. Human approval is necessary for critical actions, and the source does not specify automated failure recovery protocols beyond quality checks.
Try it
I want to set up a self-hosted OtoDock instance to automate [specific business task, e.g., weekly report generation]. My server is [Linux/Mac] with Docker installed. I have a [Claude/ChatGPT] subscription. Please guide me through: 1. Installing OtoDock via the provided script. 2. Connecting my AI engine in the Setup dashboard. 3. Creating an agent named '[Agent Name]' with the role of '[Role, e.g., Editor]'. 4. Configuring a schedule for [frequency, e.g., every Monday at 9 AM] to execute [task description]. 5. Setting up a quality check that verifies [specific output criteria]. Before enabling the schedule, run a manual test with sample data [provide sample data or file path]. Show me the output and the check result. Do not enable the automated schedule until I confirm the manual test passed. If the check fails, show me the agent's retry logic. Ask me for clarification if the task description is ambiguous.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.