Paradigm usage data

The Paradigm API can be used by company or instance administrators to track usage of the platform.

Retrieve chat sessions

import requests

base_url = "https://paradigm.lighton.ai/api/v2"

url = f"{base_url}/reporting/chat-sessions/"

headers = {
    "Authorization": f"Bearer {api_key}",
    "Content-Type": "application/json"
}

params = {
    "company_id": 2,
    "start_date": "2025-01-01",  # Required
    # "end_date": "2025-04-01"   # Optional, defaults to today
}

response = requests.get(url, headers=headers, params=params)

if response.status_code == 200:
    try:
        data = response.json()
    except requests.JSONDecodeError:
        print("Received non-JSON response:", response.text)
else:
    print("Error:", response.status_code, response.text)

Structure chat session data in a Pandas dataframe

import pandas as pd

# Flatten the structure
flattened_data = []
for entry in data["reporting"][0]["dates"]:
    flat_entry = {
        'date': entry['date'],
        'nb_sessions': entry['sessions']['sessions_count'],
        'unique_models': entry['sessions']['unique_models_count'],
        'average_session': entry['sessions']['average_session_length'],
        'user_queries': entry['chats']['user_queries_count'],
        'model_responses': entry['chats']['model_responses_count'],
        'model_responses_with_docs': entry['chats']['model_responses_with_docs_count'],
    }
    flattened_data.append(flat_entry)

# Create DataFrame
df_sessions = pd.DataFrame(flattened_data)

Going further with more endpoints

The same approach can be done with:

• the feedbacks endpoint: /reporting/chats-feedback/
• the tools usage endpoint: /reporting/tools/

Then, merging everything into the same dataframe allows a complete data analysis on the usage of the platform.