Note: Setup has two parts — (1) connecting it to the Lucidic Dashboard, and (2) installing LucidicAI in your code (below).

​
The Magic of Drop-in Observability

With just 2 lines of code, get complete visibility into your AI workflows: 
import lucidicai as lai
lai.init(providers=["openai"])  # That's it!

​
What Happens Automatically

When you add these 2 lines to any Python script:
  • Every LLM call is tracked - Automatic event creation for all API calls
  • Steps are created automatically - No manual step management needed
  • Sessions end gracefully - Automatic cleanup on script exit
  • All data flows to your dashboard - Real-time visibility
  • Works with existing code - No refactoring required

​
Installation

pip install lucidicai
Set your credentials (get them from the dashboard): 
export LUCIDIC_API_KEY=your-api-key
export LUCIDIC_AGENT_ID=your-agent-id

​
Complete Example

Here’s a real-world example showing how minimal the setup can be: 
import lucidicai as lai
from openai import OpenAI

# Just these 2 lines to add complete observability:
lai.init(providers=["openai"])

# Your existing code remains unchanged
client = OpenAI()

def analyze_text(text):
    response = client.chat.completions.create(
        model="gpt-4",
        messages=[{"role": "user", "content": f"Analyze this: {text}"}]
    )
    return response.choices[0].message.content

def summarize_text(text):
    response = client.chat.completions.create(
        model="gpt-4",
        messages=[{"role": "user", "content": f"Summarize this: {text}"}]
    )
    return response.choices[0].message.content

# Your workflow runs normally
text = "The quick brown fox jumps over the lazy dog."
analysis = analyze_text(text)
summary = summarize_text(analysis)
print(summary)

# When script exits, Lucidic automatically:
# 1. Creates steps for each distinct operation
# 2. Creates events for each LLM call
# 3. Ends all steps properly
# 4. Ends the session
# 5. Sends everything to your dashboard

​
Supported Providers

Just change the provider name in init(): 
# OpenAI
lai.init(providers=["openai"])

# Anthropic
lai.init(providers=["anthropic"])

# Multiple providers
lai.init(providers=["openai", "anthropic"])

# Other supported providers
lai.init(providers=["langchain"])       # LangChain
lai.init(providers=["pydantic_ai"])     # PydanticAI
lai.init(providers=["openai_agents"])   # OpenAI Agents

​
Want More Control?

While the 2-line setup handles everything automatically, you can take control when needed:

Use Decorators

Simplify tracking with @step and @event decorators

Session Management

Configure sessions with custom names, IDs, and metadata

Manual Steps

Create and manage steps explicitly for complex workflows

Custom Events

Track custom events beyond LLM calls

Full Examples

See complete examples with manual control and decorators

​
Next Steps

View Your Sessions

See your tracked sessions in the dashboard

Add Evaluations

Score and evaluate your agent’s performance

Run Mass Simulations

Test your agent at scale

Explore Integrations

Learn about provider-specific features

​
Minimum Requirements

  • Python 3.8+
  • Compatible with all major LLM frameworks
  • Works in local, server, or cloud environments
That’s it! You now have complete observability into your AI workflows with just 2 lines of code.