# The Open Source n8n Alternative for Agent Work That Finishes Itself

description: Kortix is the open-source n8n alternative for open-ended work: agents, company memory and 3,000+ connectors in one repo you can self-host.

Kortix is the open-source n8n alternative for teams whose workflows run only the steps they drew. When the task is research, a report, a fix or a reply that changes shape each time, Kortix runs an agent that plans its own steps, uses your tools and lands the result as a change request you review.

## Why people look for an n8n alternative

n8n is a fair-code workflow automation platform distributed under the Sustainable Use License ([docs.n8n.io](https://docs.n8n.io)). You assemble nodes on a visual canvas, add code where the built-in nodes stop, and run it self-hosted with Docker or in n8n Cloud ([n8n.io](https://n8n.io)). It connects to 1,500+ integrations and ships 9,000+ workflow templates ([n8n README](https://github.com/n8n-io/n8n)).

That model is built for work whose steps are known in advance. A workflow runs the same nodes in the same order every time, and the n8n canvas is designed to trace every step and keep a human in the loop ([n8n.io](https://n8n.io)). When the process is fixed, that is the right tool.

Most people looking for an n8n alternative want a different kind of work done. They want research that starts from an open question, a report assembled from scattered sources, a bug fixed only after the failure is reproduced, or a reply written from context. A fixed flowchart cannot choose its own next step, so it cannot finish that work.

## What Kortix is

Kortix is the open-source AI Operating System: your agents, their skills, your company memory and every connector in one git repo you own ([kortix.com/docs](https://kortix.com/docs)). Kortix runs any model with your keys and deploys self-hosted or as managed cloud, and because it is open source you can self-host it, read the code and modify it.

Six layers sit inside that repo:

- One git repo that is the company. Agents, skills, memory, connector config and triggers are files you own, so you can grep the whole setup, diff any change and roll it back.
- Every tool the company runs on. Kortix wires up 3,000+ apps plus any MCP, OpenAPI, GraphQL or HTTP API, with credentials brokered server-side and allow, ask or block on each tool call ([kortix.com](https://kortix.com)).
- Any model with your keys. Pick the model per agent, per session or per message across Claude, OpenAI, Gemini or your own OpenAI-compatible endpoint.
- A real agent harness. Each agent plans, calls tools and runs multi-step work to completion.
- A computer per session. Every session boots an isolated Linux machine on its own branch, and thousands run in parallel.
- One gate to land work. Work started from web, Slack, Teams, email, mobile, CLI or API lands as a change request a human reads as a diff ([kortix.com/docs](https://kortix.com/docs)).

## Kortix and n8n compared

Both tools connect systems and both run on infrastructure you control. n8n connects 1,500+ integrations and ships 9,000+ workflow templates ([n8n README](https://github.com/n8n-io/n8n)); Kortix connects 3,000+ apps plus any MCP or API ([kortix.com](https://kortix.com)). The difference is what does the work: n8n executes a workflow you designed, while Kortix runs configurable agents that choose their own steps and land the result through review.

| Dimension | Kortix | n8n |
|---|---|---|
| Open source | Yes, open source | Fair-code (Sustainable Use License) |
| Models | Any model with your keys: Claude, OpenAI, Gemini, your endpoint | OpenAI, Anthropic, Google, or open-source models |
| Where it runs | Self-host on laptop, VPS, VPC or on-prem; managed cloud | Self-host with Docker, or n8n Cloud; on-prem, air-gapped |
| Configuration | One git repo: agents, skills, memory, connectors as files | Visual canvas plus code nodes; git-based version control |
| Integrations | 3,000+ apps plus MCP, OpenAPI, Postman, GraphQL or HTTP | 1,500+ integrations and 9,000+ workflow templates |
| Human approval gate | Every change lands as a change request you read as a diff | Human-in-the-loop approval steps with guardrails |
| Pricing | Free 200 credits/month; Team $40 per seat/month; Enterprise self-host | Docker self-host or n8n Cloud; Business and Enterprise plans |

Kortix pricing is listed as a free tier with 200 credits each month, Team at $40 per seat each month, and an Enterprise tier that adds cloud, VPC or on-prem deployment ([kortix.com/pricing](https://kortix.com/pricing)). n8n offers a free cloud trial, a self-hosted Business plan and an Enterprise plan ([n8n.io/pricing](https://n8n.io/pricing)). For the point-by-point breakdown, read the [Kortix blog comparison](https://kortix.com/blog/kortix-vs-n8n).

## When n8n is the right call, and when you need an agent platform

n8n is a strong choice for automation whose steps are fixed: sync two systems on a schedule, route a form to the right team, transform data between APIs. The work repeats the same way, the canvas plus code covers it well, and self-hosting keeps it on your infrastructure.

Reach for an agent platform when the task is defined by an outcome rather than a sequence. Debugging a production error, drafting a report from a folder of documents, or working a lead list all require judgment about the next step, and the steps differ on every run. Kortix is built for that work, and the result still passes through a human review before it merges. The full [n8n vs Kortix head-to-head](/n8n-vs-kortix.html) breaks the choice down further, and [the other n8n alternatives worth considering](/best-n8n-alternatives.html) covers the wider field.

## Start with open-source Kortix

Hand Kortix one job and watch it land. [Try Kortix](https://kortix.com) and self-host it on your own machine or run it as managed cloud, then connect your first tool. If you already run n8n, [migrating from n8n to Kortix](/migrate-from-n8n.html) maps the concepts; to run it yourself, [self-hosting the open-source alternative](/open-source-n8n-alternative.html) has the steps. Short answers sit on the [FAQ](/faq.html).