Connecting to LangChain

Connecting to LangChain

Carlo can integrate with LangChain to monitor and enforce governance policies across your AI pipelines.

This allows you to validate prompts, responses, and model outputs before they reach the user.

Prerequisites

Before connecting Carlo to LangChain, ensure:

• Carlo SDK is installed
• LangChain is installed
• You have your Carlo API key

Step 1: Install Dependencies

Install both Carlo and LangChain.

pip install carlo-ai langchain

Step 2: Import Libraries

Import the required modules.

import carlo
from langchain.llms import OpenAI

Step 3: Initialize Carlo Guardian

Create a Guardian instance.

guardian = carlo.Guardian(
api_key=”YOUR_API_KEY”
)

Step 4: Wrap Your LangChain Model

Connect the LangChain model with Carlo governance.

llm = OpenAI()

secured_llm = guardian.wrap(llm)

Step 5: Send a Prompt

Now all prompts will pass through Carlo’s policy engine.

response = secured_llm(“Explain artificial intelligence.”)

print(response)

What Happens Now

Carlo will automatically:

• Scan prompts for unsafe input
• Enforce governance policies
• Log all requests for auditing

Next Step

Once LangChain is connected, you can enable logging and monitoring.
Continue to:
Setting up audit logs