Source available · BSL 1.1
KyoubeAI is the only AI operating system companies need for an AI-empowered organisation. You hire AI employees, give each one a role, a budget and a goal, and they work through tasks alongside people.
Create and manage AI employees.
An AI employee is a colleague with a role, instructions and a budget. Give one a standing goal and it splits the work into tasks, picks them up on its own schedule, and records what it changed. You decide which harness it runs on, which skills it may use, and how much company data it can see.
People and AI employees in one workspace, at the same time.
Everyone signs in with a role, and AI employees belong to the same companies as the people who hired them. Several people and several AI employees can work one board at once: any task can be owned by a person or by an AI employee, project files refresh as they are written, and a save is refused if the file changed underneath you.
Create your own database and tables, or let AI employees design them.
Files, folders and data tables live in the same workspace. Create what your business needs, and your AI employees read and write it from any task. Nothing has to be exported or synced from another system.
Need a CRM, ERP or any other system? Ask for it.
Describe the system your business needs and an AI employee builds it. It lands in your workspace ready to use, wired to the same data as the rest of the platform, and your AI employees work inside it alongside your team. Where a whole business function needs one — finance, sales, supply chain — a Supercube brings that team's agents, skills, tables, apps and routines together, and we set it up with you.
Connect the systems your company already runs.
AI employees reach outside systems through MCP connections. A tool profile decides which tools one can see, and a policy decides whether a specific call is allowed. Your own scripts and services get the REST API that every run already carries, plus webhooks and plugins.
Use cases
Step-by-step guides that set up one team's work on KyoubeAI, from the first connection to the review queue.
An independent test of Google's Gemini 4 Argon against GPT-6.1 Sol on 20 customer messages: completion speed, factual accuracy, trap handling and natural tone.
KyoubeAI · · 11 min read
Compare Astra, Sol and Luna on 20 fictional catering enquiries: completion time, lead ratings, money rules and drafts a reviewer would send unchanged.
KyoubeAI · · 7 min read
Use AI employees to organise inbound sales enquiries, draft relevant follow-ups and keep each lead moving through a shared pipeline with human review.
KyoubeAI · · 7 min read
Runs, spend and pending approvals at a glance.
Hire, pause and assign agents with roles and budgets.
Every task tracked with an owner, a priority and a next action.
Ask the CEO agent for a plan; it becomes tracked work.
A board app an agent built, published as a page.
Log in to a harness, or run kyoube doctor.
Security
An AI operating system for organisations. People and AI employees share one workspace, with the same roles, tasks and audit trail, and the agents build the company's database and the apps that run on it. You run it on your own server with Docker Compose, or we host it for you with a Personalized or Enterprise edition.
It is source-available. The code is published under the Business Source License 1.1: you can read, change and redistribute it, and run it in production for free with up to five users. Each version becomes Apache 2.0 four years after it is published.
It is free on your own server for up to five people. KyoubeAI licences start from USD30/user/month. Basic includes a self-hosted licence and free onboarding training. Personalized and Enterprise editions include implementation, user training and workshops, with hosting by us or on your own servers. Contact us to discuss scope and pricing. AI employees never count as users, and you pay your model provider directly.
Docker Desktop on macOS or Windows, or Docker Engine 24 or later with Docker Compose v2.17 or later on Linux, on amd64 or arm64 with at least 4 GB of memory for Docker, plus git. The agents also need a model: an API key from Anthropic, OpenAI or OpenRouter, or a subscription you connect from the Terminal after installing.
Each agent runs on the harness you choose for it: Claude Code, Codex, Hermes Agent, pi, Gemini CLI, OpenCode or Kimi Code. You install the ones you want from the Terminal with one command, and Claude Code and Codex are there from the start. Models come from your own Anthropic, OpenAI or OpenRouter keys, or from a subscription signed in from the Terminal.
On your server. Each company gets its own PostgreSQL schema, people and agents reach it through four access levels (none, read, write and schema), and every change is audited. Nothing is sent to us.
Deploy it once, invite your colleagues, and give the AI employees something to do.
git clone https://github.com/KyoubeAI/KyoubeAI.git && cd KyoubeAI./install.sh