Most people accept the first answer and move on. Power users treat it as the opening move — they dig in, push back, ask for alternatives, and explicitly ask what they missed.
One well-chosen follow-up usually unlocks more value than the original prompt.
[After: "Use Redis for session storage"]
"Why Redis over Memcached specifically? What would I lose by choosing Memcached instead for this use case?"
"Walk me through the failure scenario: Redis pod restarts under load. What happens to active sessions? Does the answer change if I enable AOF persistence?"
"We have 1,500 concurrent users now. At what scale does a single Redis setup start failing, and what's the migration path from there?"
"What's the most common mistake teams make when first setting up Redis for sessions? Give me a concrete incident story."
"Make the strongest case for NOT using Redis here. What would a skeptic say?"
[After: "This looks like a promising business model"]
"Be more critical. What's the single biggest strategic flaw that you softened or omitted because I might not want to hear it?"
"What does the most common failure mode look like for a business exactly like this? Give me a specific archetype — a real company or a named pattern — that tried this and failed."
"Name the one assumption in my model that, if wrong, makes everything else irrelevant. How could I cheaply test it in 2 weeks?"
"Write the Series A rejection email from a skeptical investor who passed on this deal."
Claude holds context across a long conversation. Reference earlier turns: 'Earlier you said X — if Y changes, does that still hold?'