# Kortix vs LangChain: an open source platform or a framework

> LangChain gives Python teams a configurable agent harness to embed in an application. Kortix is the open-source AI Operating System: it runs agents whose skills, memory, and connectors live in one git repo you own. This page compares the two layers and when to pick each.

URL: https://opensourceaiagentframework.com/kortix-vs-langchain

## Kortix and LangChain, layer by layer

Two different jobs: a platform that runs agents, a library that composes one.

| | Kortix (recommended) | LangChain |
| --- | --- | --- |
| What it is | The open-source AI Operating System a company runs and owns. | A library that composes an agent harness from model, tools, prompt, middleware. |
| Open source | Yes | Yes |
| What you own | The whole system: agents, skills, memory, and connectors in your repo. | Your application code and the harness you compose inside it. |
| Runs the agents or only defines them | Runs them: each session on its own cloud computer. | Defines them; your application runs the loop. |
| Where memory and state live | In the repo as files, versioned in the same git history. | In your app; LangGraph persistence sits under it and LangSmith holds traces. |
| Model choice | Any model with your own keys, or a ChatGPT subscription. | Any model; one interface across OpenAI, Anthropic, Google. |
| Human review of changes | Work reaches main through a change request a person reviews. | You add the review step; LangGraph can pause on state. |
| When to pick which | Company work that must be owned, reviewed, and run end to end. | An agent inside your product, with every prompt and branch yours. |

## Work lands as a diff a reviewer reads

`fix-checkout-expiry.diff`

```diff
app/checkout/charge.ts
- if (card.expMonth < now.getMonth()) {
+ if (isExpired(card, now)) {
+   return fail(402, "card_expired");
+ }
+2 -1
```

Source: https://kortix.com

## What LangChain gives a team

LangChain is a library for building an agent harness. Its own docs describe an agent as a model plus a harness, and give you `create_agent` to compose one from a model, a set of tools, a prompt, and middleware that shapes behavior. LangChain supports OpenAI, Anthropic, Google, and other providers behind one interface, so a team can switch a model without rewriting every call site. The agent machinery is built on LangGraph, which adds durable execution, persistence, and human-in-the-loop support. The application, the prompts, and the control flow around the loop stay with your engineers.

## The layer a platform adds

A harness runs one agent inside a process that you build, deploy, and watch. A platform adds the parts a company needs around that loop: where the agents and their skills live, where the memory they work from is stored, how they reach your tools, and how their output is reviewed before it counts. In Kortix those parts are files in one git repo you own. Agents and skills are markdown, memory accumulates as files, connectors are declared in `kortix.yaml`, and each session runs on its own cloud computer. Session work reaches main through a change request a person reviews, and merge is default-deny for agents.

## When to pick which

Pick LangChain when the agent is a feature of your product and your engineers want to place every prompt, tool, and branch themselves. The library stays inside your application, and LangSmith traces what happens at runtime. Pick Kortix when the agent is part of how the company works and the requirement is ownership: one repo that holds the agents, skills, memory, and connectors, work that lands as reviewed changes, and a model you can swap. [Framework or platform](/framework-or-platform) walks through the decision, and [Kortix versus LangGraph](/kortix-vs-langgraph) compares the layer under LangChain.

## Kortix and LangChain, answered

### What is the difference between LangChain and Kortix?

They sit at different layers. LangChain is a library that composes one agent harness inside your application. Kortix is the open-source AI Operating System: it runs agents on their own cloud computers and keeps agents, skills, memory, and connectors in one git repo you own.

### Is LangChain open source?

Yes. LangChain is open source. It provides `create_agent`, a configurable harness you compose from a model, tools, and middleware, and it supports OpenAI, Anthropic, Google, and more.

### Does Kortix replace LangChain?

They are different layers, so they are not competing for the same job. LangChain defines an agent inside your application. Kortix runs agents and owns the state they work from, with work landing through a change request a person reviews.

### Which should a team choose?

Choose LangChain when the agent lives inside your product and your engineers want full control of the loop. Choose Kortix when the agent is part of how the company works and the system has to be owned, reviewed, and run end to end.

## Run your agents on a platform you own

One repo for the agents, the skills, the memory, and the connectors.

[Try Kortix](https://kortix.com)