Kortix is an open-source AI operating system that runs company agents, skills, memory, and connectors from one git repo you own, with any model and self-hosting.
About
Kortix is an open-source AI operating system that lets a company run AI agents, their skills, shared memory, and tool connectors from a single git repository it owns. The problem it addresses is fragmentation: most agent tools rent out one layer of the stack while keeping configuration, credentials, and history in a vendor's database. Kortix instead stores agents, skills, memory, rules, triggers, connectors, and machine configuration as plain files in one repo, so every change can be diffed, reviewed, and rolled back like code.
At a high level, each agent session boots its own isolated Linux machine, runs work through an OpenCode-powered harness, and lands results as a change request a person approves before anything reaches main. Kortix is model-agnostic: teams can bring their own API keys from major providers, connect an existing ChatGPT subscription, or point to any OpenAI-compatible endpoint. The platform connects to more than 3,000 apps plus MCP, OpenAPI, GraphQL, and raw HTTP, and it can be self-hosted, run in a VPC, or used through a managed cloud.
Key Features
Pricing & Fees
- Pricing: A Free plan provides 500 credits per month with no credit card, one project, and the option to bring your own model key or ChatGPT subscription. The Credits tier starts at $50 a month, with a $200/month plan including 20,000 credits (about 1,830 agent-computer hours), rollover of unused credits for one month, and up to 200 projects. Enterprise pricing is custom, adding SAML SSO, SCIM, audit logs, SLA, DPA, and VPC or on-prem deployment.
- Credit economics: One credit equals $0.01 and buys about five minutes of agent-computer time; compute bills per second while a machine runs and costs nothing while stopped.
- Model costs are separate: With your own API keys, the first $1,000 a month of model usage is free and Kortix adds 5% beyond that; managed Kortix models carry a 5.5% platform fee on top of the provider's base price.
- Open source: The full stack, including the agent harness, is open source, so the harness is not a lock-in point.
Who It's For
Kortix suits engineering, operations, marketing, sales, and finance teams that want AI agents to carry out recurring work end to end rather than just answer questions. Typical users include engineering leads running background coding agents that triage errors and open fix requests, marketing teams maintaining SEO programs on a weekly cadence, and IT administrators who need governed agents with per-tool permissions, audit trails, and self-hosted deployment. It fits organizations that prefer owning their configuration as code and already pay for model access from providers like Anthropic, OpenAI, or Google.
FAQ
What is Kortix?Kortix is an open-source AI operating system that runs a company's agents, skills, memory, and connectors from one git repository, positioning itself as a self-hostable alternative to tools like Claude Cowork and ChatGPT Work.
Is there a free plan?Yes, the Free plan includes 500 credits per month with no credit card required, one project, and support for bringing your own model API key or connecting an existing ChatGPT subscription.
Which tools and models does it integrate with?It connects to more than 3,000 apps plus MCP, OpenAPI, Postman, GraphQL, and raw HTTP, and it runs models from Anthropic, OpenAI, and Google, or any OpenAI-compatible endpoint you supply.
Who owns the data and can it be self-hosted?The company's configuration and memory live in a repo the customer owns, and Kortix can be self-hosted with Docker Compose, deployed in a VPC, or run on-premises; Enterprise plans add SSO, SCIM, and audit logging.
What stops an agent from doing something unintended?Every tool call is governed by allow, ask, or block rules, connector credentials are brokered server-side and never enter the sandbox, and session work lands on main only as a change request a human reviews first.
Related Tutorials
View all →
Related Tools
View all →
Yuxi - RAG Harness for AI Agents Yuxi provides a RAG harness to enhance AI agents with retrieval-augmented generation capabilities. 
OfoxAI One API for all LLMs. Access OpenAI, Claude, Gemini & 100+ models through a unified endpoint. OpenAI SDK compatible. 
Teable - The AI Database Agent Teable is an AI-powered database agent that simplifies data management and analysis. 
MiniMax Agent: Smart Planning, Easy Travel - AI Travel Tool MiniMax is an AI chatbot that offers multimodal intelligent conversations to boost productivity and creativity. 
Zapier Zapier is a no-code AI orchestration platform that connects more than 9,000 apps so teams can build automated workflows, AI agents, and chatbots without writing code.