Two people ask the same AI the same financial question and get answers of very different quality. It’s not magic or luck: one knows how to ask and the other doesn’t. In the previous chapter we saw what AI is good for; in this one you’ll learn to get useful answers instead of generic ones. And the good news is you don’t need tricks: you need structure.
Why how you ask matters
An AI doesn’t “understand” your situation; it responds to what you give it. If you ask “how do I save more?”, it’ll return the same magazine advice you’d find anywhere: spend less, make a budget, cancel subscriptions. Correct and completely useless, because it knows nothing about you.
By contrast, if you give it context, data and a concrete task, the same AI goes from clichés to something applicable to your case. The jump in quality is huge and depends only on you. Asking well is, today, a financial skill in its own right.
The five ingredients of a good prompt
A good financial prompt usually has five parts. They don’t all have to be present every time, but the more you include, the better the answer.
1. Role. Tell it what part to play. “Act as a personal finance teacher who explains clearly and isn’t selling me anything.” This steers the tone and approach.
2. Context. Your relevant situation, without sensitive data. “I’m 35, with stable income, an emergency fund already covered, and I want to start investing for the long term.” The better the context, the more tailored the answer.
3. Data. The specific information to work on: your spending figures, the options you’re comparing, the product’s text. Without data, the AI improvises.
4. Task. Exactly what you want. Not “tell me about funds,” but “explain the differences between these two funds in fees, risk, and which profile each suits.”
5. Format. How you want the answer. “In a table,” “in plain language,” “in five points,” “with the pros and cons of each option.” The right format saves you re-reading dense answers.
Put all five together and you’ll see the leap: “Act as a finance teacher. I’m 35 and want to understand, not be recommended anything. Here are the features of these two savings accounts [data]. Compare them in a table by interest, fees and requirements, and tell me what questions I should ask before choosing.”
Ask for assumptions, doubts and sources
Here’s the nuance that separates those who use AI with judgment from those fooled by its confident tone. Always add three requests to your financial prompts:
Have it show its assumptions. “Tell me what you’re taking for granted.” To answer, an AI assumes things (a return, an inflation rate, a tax situation). If it spells them out, you can correct the ones that don’t fit your case.
Have it flag its uncertainty. “Tell me which parts of your answer are solid and which I should verify.” This forces it to drop the false confidence and mark exactly what you need to check.
Have it separate fact from opinion. “Separate the facts from the interpretations.” Useful for not swallowing an opinion dressed up as a fact.
These three sentences turn the AI from an overconfident oracle into an honest copilot that tells you where to look. It’s possibly the most valuable habit in the whole course.
Ready-to-use prompts
Here are four templates you can adapt. Remember: never paste sensitive data (account numbers, ID, passwords), which we’ll cover in depth in block 4.
Understand a product: “Act as a finance teacher. I’m pasting the terms of [product]. Explain it to me in plain language, tell me the real fees, the risks, and which profile it makes sense for. Flag what I should verify with the provider.”
Compare options: “Compare these options [data] in a table by cost, risk, liquidity and requirements. Don’t recommend one; help me see the pros and cons of each and what questions to ask myself to choose.”
Devil’s advocate: “I’m thinking of [decision]. Play devil’s advocate: give me the three strongest reasons it could be a bad idea and what risks I’m underestimating.”
Organise a budget: “Here are my spending categories from last month [anonymised data]. Group them, tell me where my money goes, which expenses look compressible and what questions I should ask myself. Don’t make up figures: use only the ones I give you.”
Common mistakes when asking
To close, the failures that most ruin a good query:
Asking for predictions or blind recommendations. “Will the market go up?”, “what should I buy?” We saw it: the AI will answer with poise and little reliability. Reframe toward understanding, not guessing.
Giving no context. Without your situation, the answer is off-the-shelf. Thirty seconds of context multiply the usefulness.
Stopping at the first answer. The best conversations are iterative: ask again, request depth, correct its assumptions. The first answer is a draft, not a verdict.
Accepting figures without verifying. Most important: if the answer includes a number, a fee or a rule you’ll use to decide, check it at the official source. Always.
Oversharing. Don’t paste sensitive personal data. You’ll learn to anonymise and protect your privacy in the final block.
With these habits, AI stops handing you clichés and starts being a real copilot. In the next block we put it to work on the most everyday and profitable task: organising your spending, building a budget, and finding the money leaks you don’t see.
