LangChain

Managed auth required

Custom LangChain and LangGraph apps

BotChap connects your website chat widget to any LangChain or LangGraph-powered HTTP endpoint. Configure the input field name, output field name, and optional API key, BotChap handles the proxy and keeps your credentials on the server.

BotChap sends only the current message. Session or conversation memory must be managed by your LangChain endpoint.
If your response doesn't match the output key, BotChap will also try response and text before falling back to the raw JSON.

Why use this integration

Use this integration when your team has built a custom chat backend with LangChain or LangGraph and needs a polished, branded website widget without rebuilding the frontend.

How it works

  1. 1
    Visitor sends a message in the BotChap widget
  2. 2
    BotChap widget sends the message to the BotChap server proxy
  3. 3
    The server POSTs the message to your LangChain endpoint using your configured input key
  4. 4
    Your LangChain app processes the message
  5. 5
    BotChap reads the reply from your configured output key and displays it

Required fields

LangChain URL
The full URL of your LangChain or LangGraph API endpoint.
API Key (optional)
If your endpoint requires authentication. Sent as Authorization: Bearer.
Input Key
The JSON field name your endpoint expects for the user message. Default: input.
Output Key
The JSON field name in the response that contains the assistant reply. Default: output.

What BotChap sends

BotChap sends a POST to your LangChain URL with your configured input key:

{
  "<your input key>": "<visitor message>"
}

Example with default keys:
{ "input": "Can BotChap qualify leads on a website?" }

Sample request

POST https://your-langchain-app.com/chat
Authorization: Bearer <your-api-key>
Content-Type: application/json

{
  "input": "Can BotChap qualify leads on a website?"
}

Sample response

{
  "output": "Yes. BotChap can capture intent, ask qualifying questions, and route lead data to your backend."
}

Best use cases

  • Custom LangChain RAG pipelines
  • LangGraph agent apps
  • In-house AI backends with HTTP APIs
  • Multi-step reasoning chains

Frequently asked questions

Does BotChap send conversation history to LangChain?
No. BotChap sends only the current message. Your LangChain endpoint must manage its own session or memory layer.
What if my output is nested?
Use a shallow output key. For deeply nested responses, use the Custom REST integration which supports a dot-notation response path.
Can I use LangServe?
Yes. LangServe exposes standard HTTP endpoints. Point BotChap at your LangServe route and configure the input/output keys to match.

Related integrations

Custom REST API
Any HTTP endpoint
Dify
Dify chat and agent apps

Ready to connect LangChain to BotChap?

Create a free widget, configure the integration, and embed it on your site in minutes.

View all integrations