How to write a good AI prompt for personal finance
Financial advice is increasingly coming from an unexpected place: your phone screen, in a conversation with an artificial intelligence. And this shift is not some far-off trend. It is already happening at massive scale.
According to a survey by Intuit Credit Karma involving more than a thousand American adults, published in September, 66% of people who have used generative AI turned to the technology for money-related questions. Among millennials and Gen Z, that number tops 80%. And it does not stop there: roughly 85% of respondents who received financial guidance from an AI actually followed through on the suggestions. That is a lot of people making personal finance decisions based on algorithm-generated answers.
But here is the part most people overlook: the quality of what you get back depends directly on the quality of what you ask. In other words, your prompt makes all the difference.
Andrew Lo, director of the Laboratory for Financial Engineering at MIT and principal investigator at the same institution’s computer science and artificial intelligence laboratory, got straight to the point during a recent presentation at Harvard University’s Griffin Graduate School of Arts and Sciences: there is a real art and science to prompt engineering. And when the topic is financial planning, that art can mean the difference between useful guidance and a response that sounds confident but is completely wrong. 😬
In the sections ahead, you will learn what AI does well, where it stumbles, how to build a prompt that actually delivers relevant answers for your financial life, and as a bonus, a reverse-engineering technique that speeds up the whole process.
What AI can actually do for your wallet
First things first: it is important to understand where artificial intelligence genuinely shines when it comes to money. Today’s language models, like OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini, were trained on massive volumes of financial content, including books, articles, reports, and discussion forums. That makes them capable of explaining complex concepts in simple terms, organizing information, and helping you think in a more structured way about your personal finances.
According to Andrew Lo himself, in an interview with CNBC, AI is generally good at providing high-level overviews of financial topics. For example, it can explain why diversifying investments matters, or why exchange-traded index funds might be better than mutual funds in certain scenarios but not others. Want to understand the difference between bonds and stocks? AI explains it with practical examples in seconds. Want to know what an emergency fund is and where to start? It gives you a clear, no-fluff step-by-step guide.
Beyond that, AI is an excellent partner for simulations and data organization. You can ask it to build a spending tracker spreadsheet, calculate how long it would take to pay off a debt with compound interest, or even draft a monthly budget based on your income. Tasks that used to require hours of research or a paid session with a specialist can now be done in minutes, with accessible language tailored to your situation. This is where financial planning starts gaining a real ally.
Another major strength is availability. AI has no office hours, does not charge per session, and will not judge you for asking something basic. That democratizes access to financial education in a way no other tool has managed before. People who never had access to a financial advisor can now have a detailed conversation about how to get out of debt, how to start investing with little money, or how to set short- and long-term goals.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Where AI stumbles and you need to watch out 👀
Despite all the benefits, artificial intelligence has serious limitations when it comes to personal finance, and ignoring them can be costly. Andrew Lo highlighted some of these issues very clearly.
The most well-known problem is what researchers call hallucination. AI can present incorrect information with absurd confidence, as if it were established fact. Lo was emphatic in describing this concern: one of the things about large language models that I find particularly troubling is that no matter what you ask it, it will always come back with an answer that sounds authoritative, even if it is not. It can cite outdated interest rates, offer recommendations based on regulations that do not apply to your situation, or suggest financial strategies that simply do not work as described. And the worst part: the text will sound so coherent that it is easy to swallow without questioning.
Another important limitation involves tax planning. It might seem counterintuitive, but AI is not great at precise numerical calculations, according to Lo. It can provide general guidance about types of tax deductions or tax rules you might consider, but asking it to run a detailed numerical analysis of your own taxes is risky. When it comes to very specific calculations tied to your personal situation, that is where you need to be very careful, the MIT researcher warned.
Then there is the issue of personal context. AI does not know your life. It does not know you have a child to raise, that you are thinking about changing jobs next year, or that you have a compromised credit history. Without that information, the recommendations it generates are generic by nature. They can be useful as a starting point, but they should never be treated as a personalized financial plan.
Brenton Harrison, a certified financial planner and founder of New Money New Problems, a virtual financial advisory firm, reinforced this point. He explained that there is so much context and complexity related to each individual’s financial situation that a human planner can draw out from their client, while someone using AI will not necessarily know they are uncovering all those nuances in their prompts. For Harrison, seeking advice from AI means you are giving it enough information to form an opinion and make a recommendation, and that is a step further than he would go with AI.
And there is the matter of accountability. A certified financial advisor is legally responsible for the guidance they provide. AI is not. It has no professional license, is not regulated by financial authorities, and will not reimburse you if a decision made based on its responses goes south. That does not mean it is useless. But it does mean you need to use it with critical awareness.
How to build a prompt that actually works for finances
Here is the practical part that changes everything. A well-crafted prompt is like a good question for an expert: the more context and clarity you provide, the more useful the answer you get back. Most people use AI in a vague way, and Harrison summed up that dynamic well: even if it is the best model in the world, if it gets a bad prompt, it can only do so much.
For financial planning, a good prompt needs at least three elements: context, objective, and constraints. Context is who you are financially speaking — your approximate income, your debts, your spending profile, your state of residence, and your tax bracket. The objective is what you want to achieve, whether that is paying off debt, building an emergency fund, retiring comfortably, or organizing a monthly budget. Constraints are the real limits of your situation — how much you can save per month, what your timeline looks like, what your risk tolerance is.
Andrew Lo gave a very concrete example during his Harvard presentation. A bad prompt for retirement would be something like: how should I retire? According to him, that is way too generic. Garbage in, garbage out.
A well-structured prompt, according to Lo, would look something like this: assume you are a fee-only fiduciary financial advisor. Here are my goals, constraints, tax bracket, state, assets, risk tolerance, and timeline. Provide, first, the baseline strategy. Second, the key assumptions. Third, the risks. Fourth, what could invalidate this plan. Fifth, what information is missing and, in particular, what you are uncertain about.
Notice the difference. In this case, the user is instructing the AI to frame its guidance as a fiduciary, which is a legal standard that requires the financial advisor to make recommendations in the client’s best interest. On top of that, the prompt already asks the AI to acknowledge its own limitations and uncertainties, which is critical for avoiding decisions based on incomplete information.
According to Lo, this process is essentially trial and error — almost like a conversation involving multiple prompts, maybe more than 20, until the user arrives at a satisfactory answer.
Reverse engineering: how to create shortcuts for future prompts
After going through that entire sequence of questions and refinements, Andrew Lo shared a technique that works as a true shortcut for future financial consultations. The idea is simple and brilliant at the same time.
When you finally arrive at a response you consider satisfactory, ask the AI one additional question: what prompt should I have used to generate the answer I was looking for?
Basically, you are asking the AI to teach you how to ask better questions. It will analyze the entire conversation, the adjustments made along the way, and synthesize an optimized prompt you can save and reuse whenever you have similar questions in the future.
According to Lo, this is a way to make your prompt engineering more efficient: you reverse-engineer the prompt by asking the AI to tell you what you should have done differently. Instead of repeating the entire refinement process with every new consultation, you start with a calibrated prompt from the jump. It is a significant time saver, especially for anyone who frequently uses AI to manage their personal finances. 🚀
Follow-up questions that make a real difference
Beyond reverse engineering, Lo recommends some additional steps that are especially important when it comes to financial matters. When you get what appears to be a solid answer, do not stop there. Always ask follow-up questions to test the limits of the guidance you received.
Some examples of questions suggested by Lo include:
- About uncertainties: what kind of information did you not have in order to make this recommendation that could lead to unreliable results?
- About confidence in the answer: how confident are you that this is the correct answer? What uncertainties do you have about the response, and what kinds of information do you not know but need in order to reach a conclusive answer?
This way, you can extract the range of uncertainty behind the AI’s response, understanding not just what it recommends but also where it might be wrong or where data is missing for a more robust conclusion.
Along the same lines, Harrison, the financial planner, recommends demanding that the AI list its sources. You can even instruct it to limit its sources to those that meet certain reliability criteria. According to Harrison, if you do not require it to verify sources, it will give you an opinion, and opinion is not what we are looking for in this context.
Quick tips for using AI more intelligently 💡
Mastering the use of artificial intelligence for personal finance does not require any advanced technical knowledge. It is more about mindset than skill. Below are some practices that make a real difference in day-to-day use:
- Be specific with context: the more relevant information you include in the prompt, the more personalized the response will be. Include details like income range, goals, timeline, risk tolerance, and real-world limitations.
- Ask for sources and verification: whenever the AI presents numbers, rates, or regulations, ask it to indicate where that information comes from and double-check it yourself through official sources.
- Use AI to learn, not just to get answers: ask it to explain the reasoning behind its recommendations. This helps you understand the financial logic and make more independent decisions going forward.
- Iterate the conversation: do not limit yourself to a single question. Use the AI’s response as a starting point and dig deeper with new questions. According to Lo, the process can involve more than 20 prompts before reaching a satisfactory result.
- Define the AI’s role right from the start: instructing the AI to behave as a fiduciary advisor or a financial educator completely changes the tone and quality of the responses.
- Always question the certainty of the answer: ask the AI what it does not know, what could invalidate the plan, and what data is missing for more precise guidance.
- Save your best prompts: use the reverse-engineering technique to build a small library of optimized prompts that can be reused in future consultations.
- Combine with trusted sources: use AI alongside official portals and regulated financial education platforms. AI organizes and explains, but official sources validate.
AI and finance: a powerful tool with important caveats
Artificial intelligence is not going to replace a good certified financial planner, especially in more complex situations involving personal, legal, and tax nuances. Andrew Lo was clear in saying that people should use AI for financial planning, but that how they use it is what truly matters.
Brenton Harrison complemented that view by noting that there is a difference between using AI as a learning tool and using it as a source of personalized advice. The first approach is healthy and productive. The second requires caution, because it assumes you are providing enough information for it to form a qualified opinion — something that does not always happen in practice.
AI is already, today, a powerful tool for anyone who wants to better understand their own money, build healthier financial habits, and make more informed everyday decisions. The secret lies in knowing how to talk to it, understanding its limits, and always, always verifying what it delivers. And now you have a solid path to get started. 🚀
