What is Gemini Deep Research and how does it work as a managed agent?

from google import genai
client = genai.Client()

# Deep Research agent - model ID may vary; check current docs
# Typically accessed via AI Studio or the interactions API:
interaction = client.interactions.create(
    model="gemini-deep-research-preview",
    input="""Conduct comprehensive research on the performance
    characteristics of Python async frameworks in 2026:
    FastAPI, Starlette, Litestar, and Blacksheep.
    Compare throughput, latency, ecosystem maturity,
    and real-world adoption. Produce a structured report
    with data sources cited.""",
    background=True,  # Long research tasks run in background
)
print(f"Research started: {interaction.id}")

# Poll for completion (research may take 5-30 minutes):
import time
while True:
    result = client.interactions.retrieve(interaction.id)
    print(f"Status: {result.status}")
    if result.status == "completed":
        print(result.output_text)  # full research report
        break
    elif result.status == "failed":
        print("Research failed")
        break
    time.sleep(60)