Skip to main content

​
Send Observations From Your Application

To track how your applications are using data, you need to send observations to Arcbeam using an API key.

​
Create API Key

Before you can send observations, you need to generate an API key:
  1. Navigate to the settings page by clicking on the User icon on the top right of the screen
  2. Click on the API Keys menu item on the left hand side
  3. Generate an API key from the Create API Key button
API key generation screen

​
Installation

To capture usage, your application can use our LangChain connector to send data to the Arcbeam API whenever a query is processed. 
pip install arcbeam-langchain-connector
For more information, visit the arcbeam-langchain-connector on PyPI.

​
Minimal Python Helper

Here’s a basic example of how to use the connector: 
# Basic LLM call with tracing
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage
from arcbeam_langchain_connector import ArcbeamHandler

handler = ArcbeamHandler(
    base_url="YOUR_ARCBEAM_URL",
    api_key="YOUR_ARCBEAM_API_KEY"
)

llm = ChatOpenAI(callbacks=[handler])
response = llm.invoke([HumanMessage(content="What is the weather like today?")])
print(response.content)

# For chains and agents, pass handler in config
from langchain_core.prompts import ChatPromptTemplate

prompt = ChatPromptTemplate.from_template("Explain {topic} in two sentences.")
chain = prompt | llm

result = chain.invoke(
    {"topic": "AI monitoring"},
    config={"callbacks": [handler]}
)
Replace YOUR_ARCBEAM_URL with your Arcbeam deployment URL (e.g., http://localhost:3000) and YOUR_ARCBEAM_API_KEY with the API key you generated.

​
View Results in the UI

Return to the Projects page and select your project. You’ll see each query recorded along with:
  • Retrieved chunks
  • Model used
  • Input and output text
Query results in the UI
During the private beta, applications may experience increased latency because data enrichment is performed inline rather than in the background.

​
Next Steps

  • Configure projects to organize your monitoring and apply policies
  • Review observations and refine how your datasets are used