---
sidebar_label: 'Overview'
---

import {CodeEditor} from '@site/src/components/CodeEditor/CodeEditor';
import exampleUrlFileAiChatBot from './examples/aiChatBot-default';

# LangChain Adapter

`NLUX` offers integration with LangChain through LangServe.

## About LangChain and LangServe

[LangChain](https://www.langchain.com/) is a popular framework for building backend services powered by large language
models. It enables applications that are context-aware. It allows connecting a language model to sources of context
such as databases, APIs, and other services. It also relies on language models to reason about the context.

[LangServe](https://www.langchain.com/langserve) is a library that comes with LangChain, and that allows developers to
build Restful APIs that can be used to interact with language models. A LangServe API provides a set of pre-defined
endpoints and an input/out schema that can be used to interact with the language model.

:::note LangServe vs Custom APIs
The `NLUX` LangChain adapter supports the **[standard LangServe endpoints](https://python.langchain.com/docs/langserve#endpoints)**.<br />
If you have created custom APIs powered by LangChain, *that do not use LangServe*, we recommend using
`NLUX`'s [Custom Adapters](/learn/adapters/custom-adapters) feature to create adapters specific to your APIs.
:::


## Supported Features

The `NLUX` LangServe adapter can do the following:

* Offers a way to **send a user prompt to a LangServe API** via `/invoke` or `/stream` endpoints.
* **Handles the API responses** and displays them in the AI chat UI (single responses and streamed text).
* Offers a way for developers to **customize the payload sent and received** to/from the LangServe API.
* Reads the LangServe API **input schema** and attempts to construct a request payload that matches it.

It's also important to note what `NLUX` LangServe adapter does not do:

* Does not handle `/batch` or `/stream_log` endpoints.
* The generation of request payload based on input schema is limited to schemas with simple types
(string, or object with a single string property). Of complex schemas, `inputPreProcessor` should be
used to construct the payload.

If you have specific requirements that are not covered by the adapter, please submit a feature request
[here](https://github.com/nluxai/nlux/discussions) and we will consider adding it to the roadmap
if enough users request it.

## Example: LangServe Adapter With Default Settings

The example below shows how to use the `NLUX` LangServe adapter to connect to a LangServe API.

* LangServe Runnable URL used: `https://pynlux.api.nlux.ai/pirate-speak`
* Data transfor used: `stream` (default) ― Response is streamed to the UI as it's being generated.

<CodeEditor
    files={{
        'App.tsx': exampleUrlFileAiChatBot,
    }}
    editorHeight={380}
    simulatedPrompt={'What\'s the best human food treats to feed a pirate parrot to keep him happy during long voyages at sea?'}
/>