Kortix vs Open WebUI: chat with your models, or an open-source company system?

Open WebUI is the self-hosted interface teams use to talk to their own models; Kortix is the open-source AI Management System they adopt when work has to leave the chat window. Both are self-hostable and point at the models you choose. Open WebUI fits a team that wants a private, model-agnostic chat surface, while Kortix fits a company that needs agent work, shared memory and connectors to live in a repo it owns.

What Open WebUI is great at

Open WebUI is a self-hosted AI platform built to run entirely offline, and it connects to Ollama or any OpenAI-compatible API. (Open WebUI on GitHub)

• One command to a working chat. The quick start is a single docker run against a published image, with Python and Kubernetes paths for larger deployments. (quick start)
• Retrieval built in. Local RAG runs on nine vector databases with hybrid BM25 and vector search plus reranking, and you load documents into a chat with # or pull them from the library.
• Search, images and notes. Web search spans dozens of providers, image generation and editing run through several engines, and Notes and Channels sit beside the conversation.
• Enterprise identity and scale. LDAP, Active Directory, SSO and SCIM 2.0 provisioning are supported, and the application layer stays stateless over shared databases and storage while Redis coordinates sessions. (enterprise overview)

Where a chat interface stops

Open WebUI is built around the conversation, so its state and execution stay close to the chat.

• Memory is personal. Open WebUI remembers facts about you across conversations, so what it learns belongs to one person's chat history and is not shared with the company.
• Agent execution accompanies the chat. Open Terminal gives an agent a terminal and filesystem inside the chat, and per-user isolated Terminals requires an Enterprise licence.
• Configuration lives in the product. Models, tools and knowledge bases are managed through the deployment's database and admin panel and are not versioned as files.
• Work ends where the chat ends. A finished answer is a message in a thread, and nothing about it updates the company's systems by default.

The company is one git repo

Kortix stores the agents, the skills they share, the company's memory and every connector config as plain files in one git repository you own. You can grep the whole company, diff any change and roll it back, because the configuration is version controlled and does not live inside a product database. (Kortix on GitHub)

A new teammate clones the Kortix repo and gets the whole company; any change to an agent or a skill arrives as a diff you review. Agents are written in Markdown and the runtime config lives in kortix.yaml, so the system reads like documentation and reviews like code. Kortix runs on any model with your own keys.

Work lands as a change request

Every Kortix session runs on its own isolated Linux computer, on a branch named after the session. The agent can install, run and break anything, and only what it commits survives. Work reaches main through a change request a human approves, and thousands of sessions run in parallel on the same configuration.

Kortix connects to 3,000+ apps plus MCP, OpenAPI, GraphQL and raw HTTP APIs, with connector credentials brokered server-side so the raw key never enters the machine.

Self-hosting Kortix is one Docker Compose stack that holds the frontend, the API, the LLM gateway and the Supabase distribution, with agent sessions on a separate sandbox provider (Daytona by default; Platinum and E2B also supported). Install the CLI with curl -fsSL https://kortix.com/install | bash, then run kortix self-host init --domain kortix.example.com and kortix self-host start. (self-hosting docs)

Where Open WebUI is the better pick

For a person or a chat-first team, Open WebUI is the better pick. It is built for conversation and goes deep there: one Docker command to start, nine vector databases behind retrieval, dozens of search providers, and LDAP, SSO and SCIM when the team grows. If your work lives in conversations and you want a private, model-agnostic chat surface, Open WebUI gets you there with less setup than Kortix.

Side by side

Open WebUI ships under the Open WebUI License, a permissive licence with a branding clause: the Open WebUI name has to stay visible unless a deployment covers 50 or fewer users in a 30-day window, holds written permission, or carries an enterprise licence. (Open WebUI License) Kortix is open source (Elastic License 2.0): self-host, read and modify the code.

When to pick which

Pick Kortix when you want the company's agents, skills, memory and connector rules in one repo you own, each session on its own isolated Linux computer, and every change landing as a diff a human approves. You can run any model with your own keys, and self-host on a laptop, a VPS, your VPC or on-prem, or use managed cloud. The free tier costs nothing to start, and Team runs $40 per seat monthly, checked October 2026. (Kortix pricing)

Pick Open WebUI when the work lives in conversations: a fast, private, model-agnostic chat surface for you and your team, with retrieval over nine vector databases and enterprise identity when you need it.

Read how the open-source AI agent platform category fits together, what to weigh in a self-hosted AI workspace, and how Kortix lines up against OpenWork.

Free to start, free to self-host. Get started with Kortix.