Find the traces that matter.

Add one decorator. Wrap your client. That's it.

from openai import OpenAI
from playgent import record, wrap_openai

client = wrap_openai(OpenAI())

@record
def inference(user_input: str):
    response = client.responses.create(
        model="gpt-4",
        input=[{"role": "user", "content": user_input}]
    )
    return response.output_text

Playgent monitors your agent in production and automatically flags sessions where users struggle or agents fail to accommodate requests.

Automatically identify and flag incidents where your agent struggles or fails to meet user needs

Track tokens, costs, latency, and performance metrics with built-in observability