Here’s the one thing I do every single time I get advice from an LLM, whether it’s Claude, ChatGPT, or Gemini:

I ask “What are the trade-offs?”

That’s it. That’s the trick.

Nothing in life is free. Every choice has a cost, a downside, a “yeah but.” The problem is that LLMs, by default, tend to give you a clean, confident answer. They present a solution as if it’s the obvious right call. And sometimes it is. But you’re missing context if you don’t ask what you’re giving up.

Say I ask an LLM how to structure a new data pipeline. It’ll give me an architecture that looks great. Clean. Elegant, even. But if I follow up with “What are the trade-offs of this approach?”, suddenly I learn that it adds operational complexity, or that it doesn’t scale well past a certain volume, or that it couples two systems that I might want to keep independent.

That follow-up question turns a recommendation into a decision framework. Instead of blindly following advice, now I have the full picture: what I gain and what I give up. I can weigh the trade-offs myself based on context only I have.

The thing is, the LLM knows about these trade-offs. It’s not hiding them maliciously. It’s just optimizing for giving you a direct, helpful answer. And a direct answer means picking a side. So you have to explicitly ask for the other side.

I do this for everything: technical decisions, career advice, how to phrase a difficult email, you name it. It’s become reflexive at this point.

Next time you get a recommendation from an LLM, try it. Ask: “What are the trade-offs?” You might be surprised how much more useful the answer becomes.