How do you implement multi-turn conversations in the Gemini API?

from google import genai
from google.genai import types

client = genai.Client()

# Approach 1: Manual history with generateContent (legacy)
history = []
while True:
    user_input = input("You: ")
    history.append(types.Content(role="user", parts=[types.Part(text=user_input)]))

    response = client.models.generate_content(
        model="gemini-3.5-flash",
        contents=history,
    )
    assistant_reply = response.candidates[0].content
    history.append(assistant_reply)   # add model turn to history
    print(f"Gemini: {assistant_reply.parts[0].text}")

# Approach 2: Interactions API with previous_interaction_id (recommended)
# First turn:
first = client.interactions.create(
    model="gemini-3.5-flash",
    input="I am building a task management app in Python.",
)

# Second turn - model remembers the first:
second = client.interactions.create(
    model="gemini-3.5-flash",
    input="What database would you recommend for my app?",
    previous_interaction_id=first.id,  # server retrieves first turn context
)

# Third turn:
third = client.interactions.create(
    model="gemini-3.5-flash",
    input="Show me the schema for that database.",
    previous_interaction_id=second.id,
)
print(third.output_text)