How to beginner · 3 min read

How to use LangSmith without LangChain

Q: How to use LangSmith without LangChain

Use the LangSmith Python SDK directly by installing langsmith, initializing a Client with your API key, and manually tracing your AI calls with the @traceable decorator or client.trace() method. This approach enables LangSmith's observability features without requiring LangChain.

Quick answer

Use the LangSmith Python SDK directly by installing langsmith, initializing a Client with your API key, and manually tracing your AI calls with the @traceable decorator or client.trace() method. This approach enables LangSmith's observability features without requiring LangChain.

PREREQUISITES

Python 3.8+
LangSmith API key
pip install langsmith

Setup

Install the langsmith package and set your LangSmith API key as an environment variable.

Install LangSmith SDK: pip install langsmith
Set environment variable: export LANGSMITH_API_KEY=your_api_key (Linux/macOS) or set LANGSMITH_API_KEY=your_api_key (Windows)

bash

pip install langsmith

Step by step

Initialize the LangSmith Client and use the @traceable decorator to trace a simple function that calls an AI model. This example shows how to manually trace calls without LangChain.

python

import os
from langsmith import Client, traceable

# Initialize LangSmith client with API key from environment
client = Client(api_key=os.environ["LANGSMITH_API_KEY"])

@traceable(client=client)
def simple_ai_call(prompt: str) -> str:
    # Simulate an AI call (replace with your actual AI call)
    response = f"Response to: {prompt}"
    return response

if __name__ == "__main__":
    result = simple_ai_call("Hello LangSmith without LangChain")
    print("AI call result:", result)

output

AI call result: Response to: Hello LangSmith without LangChain

Common variations

You can also use client.trace() as a context manager for manual tracing or integrate LangSmith with other AI SDKs by wrapping calls. Async tracing is supported similarly with async functions.

python

import asyncio
from langsmith import Client

client = Client(api_key=os.environ["LANGSMITH_API_KEY"])

async def async_ai_call(prompt: str) -> str:
    # Simulate async AI call
    await asyncio.sleep(0.1)
    return f"Async response to: {prompt}"

async def main():
    async with client.trace("async_call") as trace:
        response = await async_ai_call("Async LangSmith tracing")
        trace.add_output(response)
        print("Async AI call result:", response)

if __name__ == "__main__":
    asyncio.run(main())

output

Async AI call result: Async response to: Async LangSmith tracing

Troubleshooting

If you see authentication errors, verify your LANGSMITH_API_KEY environment variable is set correctly.
Ensure network connectivity to LangSmith's API endpoint.
For missing imports, confirm langsmith is installed and up to date.

✅

Key Takeaways

Use the LangSmith Python SDK directly by initializing Client with your API key.
Trace AI calls manually with the @traceable decorator or client.trace() context manager.
LangSmith works independently of LangChain for observability and tracing.
Async and sync tracing are both supported in the standalone SDK.
Always set LANGSMITH_API_KEY in your environment for authentication.

Verified 2026-04

Verify ↗