Python SDK platformclient — retry logic for intermittent 503 errors

I keep getting 503 errors from the Python SDK and I don’t know how to handle them.

# My script crashes every time
try:
    result = api.get_users(page_size=100)
except Exception as e:
    print(f'Error: {e}')  # ApiException: 503

Sorry for the basic question, but should I just add a retry loop? How many times should I retry before giving up?

I don’t understand what a ‘503 error’ means. Can someone explain it without using technical jargon?

My agents keep telling me their screens freeze and calls drop. Is a 503 the reason? How do I make it stop?

From an AI strategy perspective, intermittent 503 errors directly impact Agent Assist reliability.

If the Agent Assist knowledge query returns a 503, the agent sees no article suggestions for that interaction. Our analysis shows that 503 errors correlate with platform maintenance windows. Implement retry with exponential backoff: wait 1s, 2s, 4s, max 3 retries.