OtoDock 1.6.0 – Self‑hosted AI Agent Platform for Company‑wide Automation

TL;DR – What OtoDock 1.6.0 delivers and why it matters

OtoDock 1.6.0 provides a self‑hosted, multi‑tenant "company OS" where AI agents powered by Claude Code or OpenAI Codex can be created, organized into departments, and run autonomously on your own Anthropic/OpenAI subscriptions or local models. The platform bundles security sandboxes, role‑based access, scheduling, phone‑call capability, and a full‑screen dashboard, making it possible to automate internal workflows without handing data to a third‑party SaaS.

Core proposition – an AI‑powered operating system for the enterprise

OtoDock treats each AI agent as a digital employee with six configurable components – Persona, Memory, Workspace, Knowledge, Skills, and Tools – all editable through a web UI.
Agents can be shared in four modes (personal only, personal + shared, shared + personal, shared only), giving fine‑grained control over who sees which workspace and history.
Roles are layered: platform‑level Admin/Creator/Member and per‑agent Manager/Editor/Viewer, enabling precise delegation.
The platform runs on your own server (Docker‑based, minimum 4 GB RAM) and can pair with any laptop, workstation or PC via a one‑line install, keeping files in sync and requiring no inbound ports.

Architecture at a glance

dashboard/   – React UI (chat, agents, files, admin)
proxy/       – FastAPI core (sessions, security, scheduling, sandbox)
mcps/        – MCP tool servers (files, memory, meetings, etc.)
audio/       – Speech‑to‑text / text‑to‑speech package
phone/       – Twilio / Asterisk / FreePBX telephony daemon
satellite/   – Remote‑machine agent for paired devices
scripts/     – Install, backup, maintenance tooling

Each agent runs as a Claude Code or Codex process inside a kernel sandbox with isolated mount and process namespaces.
Network access is blocked by default; admins grant scoped service access per agent.
PostgreSQL stores persistent state, including editable memory files per user/agent.

Feature highlights that differentiate OtoDock

Feature	What it does
Live dashboards	Real‑time UI for each agent, showing workspace, tools, and notifications.
Document editing	Agents can create and edit Word, Excel, PowerPoint, PDF files directly in the chat.
Artifacts & meetings	Agents render interactive UI (charts, budgets) and can hold live meetings with other agents, streaming results.
Terminal integration	Claude Code or Codex terminals open inside the dashboard with pre‑loaded tools.
Phone integration	Agents can receive and place calls via Twilio or self‑hosted Asterisk/FreePBX.
Remote‑machine pairing	One‑line client syncs a laptop/PC to the server; the same agent runs locally with full device access.
Security‑first defaults	Kernel sandbox, network isolation, encrypted credentials, SSO/OIDC, 2FA, per‑user cost budgets.
Extensible MCP catalog	Community‑maintained agents, tools, and skills install with a single click (e.g., Notion, Home Assistant, Prometheus).

Community and licensing

Fair Source License (FSL‑1.1‑Apache‑2.0) – free to run, modify, and redistribute for any purpose except competing commercially; automatically re‑licenses to Apache 2.0 after two years.
Free for up to 5 users; larger teams purchase seats per the pricing page.
Documentation, community agents, MCPs, and skills are hosted on GitHub under the OtoDock organization.

Real‑world considerations raised by early adopters

"Looks great. I didn't like the paperclip system with my main issue being that agents would go over daily budgets all too often. Looking forward to giving this a try." – nerdsnipe (HN comment)

"I work on a realtime collaboration platform and have built something similar. I'm turned off by the 3‑D visualizations and AI‑heavy marketing, and the mobile site is unusable on Firefox/Android." – dtkav (HN comment)

"If you organize agents by department, who verifies an agent's 'completion' report? In my experience, reviews using the same model rarely catch its own mistakes." – NimadFlow (HN comment)

These comments highlight three practical concerns:

Cost control – OtoDock includes per‑user and per‑agent budgeting, but teams should monitor token usage and set sensible limits.
UX maturity – The current UI emphasizes rich visualizations; a streamlined mobile experience may be needed for field users.
Verification workflow – The platform’s role model (Manager vs. Viewer) allows a human Manager to audit an agent’s output, but integrating deterministic validation tools (e.g., test harnesses, checksum verification) is advisable for critical tasks.

Quick start – getting OtoDock running in minutes

mkdir otodock && cd otodock
curl -fsSLO https://raw.githubusercontent.com/OtoDock/oto-dock/main/scripts/install.sh
bash install.sh

The installer checks Docker, creates a .env with a generated DB password, and pulls a release‑pinned docker-compose.yml.
After installation, browse to http://localhost:8400, complete the wizard, and the platform automatically creates a Personal Assistant agent for the owner.
Connect your Anthropic (Claude Code) or OpenAI (Codex) API keys under Setup → AI Engines.
From the Agents page you can launch the assistant, add community agents, or build custom ones.

How OtoDock fits into the AI‑infra landscape

Compared to Cloudflare OS – Both aim to provide an internal OS‑like layer for automation. OtoDock differs by being self‑hosted, multi‑tenant by design, and engine‑agnostic (Claude Code, Codex, or local models). It does not rely on Cloudflare’s edge network and gives full control over data residency and security policies.

Bottom line

OtoDock 1.6.0 delivers a comprehensive, self‑hosted platform that lets companies turn LLMs into persistent, role‑aware digital employees. Its sandboxed execution, rich UI, and extensible tool catalog make it a viable alternative to cloud‑only AI automation services, provided teams address budgeting, verification, and mobile UX concerns.