- Open source
- Yes
- What it is
- The open-source AI Operating System: one repo you own.
- Fits best when
- A team wants its agents as files it owns.
- Where agents run
- Kortix Cloud, your VPC, or your own on-prem network.
- Models
- Any model provider with your own keys.
- Where your configuration lives
- Files in a git repo you own.
- Team access
- Members, groups and roles; one sandbox per session.
- Cost to self-host
- Self-host is free; you pay your own model and compute.
Self-host an open-source AgenticSeek alternative
AgenticSeek runs a local-first agent on one machine. If your team needs the same idea self-hosted for several people, Kortix is the open-source AI Operating System: agents, skills, memory and connectors in one git repo you own.
From install to a server of your own.
Real Kortix commands, from the README and the self-hosting docs.
$ curl -fsSL https://kortix.com/install | bash$ kortix self-host init --domain kortix.example.com --admin-email you@example.com$ kortix self-host start$ kortix self-host doctorWhat self-host means for a local agent
AgenticSeek is a local-first AI assistant that runs on one machine (AgenticSeek on GitHub). AgenticSeek is voice-enabled: it browses the web, writes code and plans tasks, and it keeps your files, conversations and searches on the device. You install Python, Docker and a local model server, set is_local and a provider in config.ini, and start it. For one person, the software, the model and the data all sit on hardware you control.
The same shape does not extend to a team. AgenticSeek keeps its settings in config.ini and .env on that machine, and the model runs next to the files you point it at. Several people sharing one set of agents, with a separate environment for each task and one place the finished work goes for review, need a platform whose unit is the team rather than the machine.
What Kortix self-hosts
Kortix is the open-source AI Operating System: your agents, their skills, your company memory, and every connector live in one git repo you own. Kortix is open source. Read it, fork it, audit it. It supports three hosting shapes. Run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free.
A self-hosted Kortix is one Docker Compose stack on a Linux server you control (self-hosting docs). The stack runs the web app, the API and the LLM gateway. Agent sessions run on a cloud sandbox provider instead of that server, so the instance stays reachable for those sandboxes to call back. The first run asks for an admin email, a sandbox provider key and a preview domain. GitHub connects later in the dashboard under Settings → Git, and models are added in the app with your own keys.
Self-hosting keeps the repo, the model keys and the data on infrastructure you control. Because agents, skills, memory, connectors and triggers are files, the same repo runs on a laptop and on the server, and git records every change. Any model provider with your own keys, so the model choice stays yours. One isolated sandbox per session, so two people's work stays apart on a single configuration. Session work reaches main through a change request, so a person reviews what an agent changed.
A self-hosting checklist for a team
Before the first kortix self-host start:
- A Linux server with a public IP. The bootstrap installs Docker and the Compose plugin.
- DNS A or AAAA records for your domain and its API subdomain, plus a wildcard record for previews.
- Ports 80 and 443 open inbound.
- A sandbox provider key. Without one, sessions do not start.
- A model provider key, added in the app after the first sign-in.
- The instance reachable from the internet, so sandbox sessions can call back to it.
Which option fits a team
The comparison table on this page sets Kortix against AgenticSeek, OpenHands and Open WebUI on the parts that decide a self-hosted team deployment. AgenticSeek is at its strongest for one person on one machine. OpenHands runs coding agents for an engineering team, and Open WebUI is a chat and knowledge hub over models. Kortix is the option built for a team that wants its agents, memory and connectors in one repo it owns.
Self-host Kortix or keep a single-machine agent
Each column reflects the project's own README, checked October 2026.
- Open source
- Yes
- What it is
- A local, voice-enabled assistant that browses and codes.
- Fits best when
- One person works on one machine.
- Where agents run
- Your own machine, or a remote LLM server.
- Models
- Local Ollama or LM Studio; optional cloud keys.
- Where your configuration lives
- config.ini and .env on the machine.
- Team access
- A personal assistant for one machine.
- Cost to self-host
- Free; you pay your own hardware and power.
- Open source
- Yes
- What it is
- A self-hosted console for coding agents.
- Fits best when
- An engineering team runs coding agents.
- Where agents run
- Your laptop, a dedicated computer, or a cloud server.
- Models
- Any LLM you bring.
- Where your configuration lives
- In the app; each conversation mounts its own workspace.
- Team access
- A team can share one Agent Server.
- Cost to self-host
- Free; cloud and enterprise plans are optional.
- Open source
- Yes
- What it is
- A self-hosted interface for local and hosted models.
- Fits best when
- A team wants a shared chat and knowledge hub.
- Where agents run
- Your server, on Docker, Kubernetes, or a plain process.
- Models
- Ollama and any OpenAI-compatible API.
- Where your configuration lives
- In its own database, SQLite or PostgreSQL.
- Team access
- Roles, groups and permissions; SSO and SCIM.
- Cost to self-host
- Free; some capabilities are Enterprise only.
Self-hosting questions
Can I self-host Kortix on my own servers?
Yes. Run it on Kortix Cloud, in your VPC, or on your own on-prem network, and self-host is free. The CLI installs one Docker Compose stack on a Linux server you control.
What does self-hosting Kortix cost?
Self-host is free. You pay only the sandbox provider and the model keys you connect to the instance.
Does a self-hosted instance work fully offline?
No. kortix self-host start pulls images from docker.io, and the instance must be reachable so sandbox sessions can call back to it. You run it on your own network, but it needs those connections.
How is Kortix different from running AgenticSeek on one machine?
AgenticSeek is a local assistant for a single machine. Kortix keeps your agents, skills, memory and connectors in one git repo you own, runs each session in its own isolated sandbox, and lets several people work from the same configuration.