What’s the difference?
The way they’re asking.
A prompt isn’t just a question - it’s a set of instructions. Think of it as programming the AI to give you exactly what you need instead of just… whatever it feels like spitting out.
Let me show you how this works, specifically for money stuff.
What actually is a prompt? #
A prompt is basically your way of telling an AI model what you want. But there’s a huge difference between asking like you’re Googling something and actually giving the AI enough info to help you.
Bad version: “Help me budget”
Good version: “Act as a certified financial planner. I make $85,000/year, my fixed expenses are $3,500/month. Create a detailed budget using the 50/30/20 rule with specific dollar amounts and savings goals.”
See what happened there? The second one gives the AI everything it needs. Who to be. What numbers to work with. How to format the answer.
Why this matters for your money #
Financial decisions compound. You know this. A mediocre budget over ten years costs you thousands. A poorly thought-out investment strategy? Tens of thousands left on the table.
If you can get AI to give you actually useful financial advice - like, really tailored to your situation - you’re basically getting a financial advisor for free. But only if you know how to ask.
The four things every good prompt needs #
Alright, there’s four pieces to this puzzle. Get these right and you’ll get better advice than most people pay hundreds for.
1. Give it a persona #
This is where you tell the AI what kind of expert you need it to be. And this totally changes the answer you get.
Here’s what I mean:
Generic: “How should I invest?”
With a persona: “You’re a fee-only fiduciary financial advisor who’s helped middle-income people reach financial independence for 20 years. How should I invest?”
That second version? It’s telling the AI to think like someone who has a legal duty to act in YOUR best interest. Someone who doesn’t get paid commissions. Someone who knows the path to FI.
More examples you can use:
- “You’re a tax specialist who helps high earners optimize their tax strategy…”
- “You’re a retirement specialist who helps people in their 40s catch up on savings…”
- “You’re a behavioral economist who gets why people struggle with spending…”
- “You’re a debt elimination coach who’s helped thousands pay off six-figure debts…”
Give your AI a personality. Seriously, it makes a massive difference.
2. Load it up with context #
Context is the details about YOUR situation. This is what turns generic advice into something you can actually use.
No context: “Give me investment ideas under $10,000”
Loaded with context: “I’m 32, have $10,000 to invest. Already maxing out my tax-advantaged retirement accounts. Emergency fund’s covered. I’ve got moderate risk tolerance, 30-year time horizon. Want to diversify outside retirement accounts. Give me five specific options with pros and cons.”
The second one tells the AI:
- How old you are
- What you’ve already done
- How much risk you’re okay with
- What timeframe you’re working with
- Exactly what format you want
Always be contexting!
Throw in everything that’s relevant:
- Your income and expenses
- Current debts and assets
- Time horizons (when do you need this money?)
- How much risk you can stomach
- Family situation
- Career stuff
- Your actual goals
The more you give it, the better it gets.
3. Tell it how to format the answer #
This is where you get specific about how you want the information back. Format, length, tone. All of it.
Format example:
“Present the budget breakdown as:
- A table with categories, amounts, and percentages
- A ranked list of where I can cut spending
- Action steps I can do this week”
Length example:
“Keep it under 300 words, but give me a detailed spreadsheet format for the calculations.”
Tone example:
“Be professional but encouraging. I need straight talk, not sugar-coating. But also don’t make me feel bad about my mistakes.”
For financial stuff, you can ask for:
- Spreadsheet-ready formats
- Comparison tables
- Decision matrices
- Priority rankings
- Step-by-step plans
Whatever makes it easiest for YOU to use.
4. Show it examples (few-shot prompting) #
This one’s sneaky powerful. You literally show the AI exactly what you want by giving it an example.
Like this:
“I want you to analyze investment options. Format your response EXACTLY like this:
Investment Option: Global Stock Market Index Fund Risk Level: Moderate (6/10) Minimum Investment: $1,000 Expected Return: 7-10% annually Pros:
- Low expense ratio (typically 0.05-0.20%)
- Broad diversification across thousands of stocks
- Strong long-term historical performance Cons:
- No downside protection in bear markets
- Value fluctuates with market conditions Best For: Long-term investors (10+ years) comfortable with market swings
Now analyze these three investments using the same format: [your list]”
The AI will copy your structure exactly. Makes comparing options super easy.
Advanced stuff (Chain of Thought and Tree of Thoughts) #
Okay so once you’ve got the basics down, there’s some advanced techniques that’ll blow your mind.
Chain of Thought (making it show its work) #
This is where you tell the AI to think step-by-step before answering. It’s like when your math teacher made you show your work, except here it actually helps.
Example for retirement planning:
“I’m 35 with $50,000 saved for retirement. Want to retire at 55 with $2 million. Before you give me a plan, think through step-by-step:
- How much do I need to save monthly to hit $2 million by 55?
- What investment return assumptions are we using? Are they realistic?
- What could derail this plan? (Market crashes, job loss, inflation)
- What adjustments could I make if I fall behind?
Show all your calculations and reasoning, then give me the recommendation.”
Why this works:
- The AI catches its own mistakes
- You see the reasoning (builds trust)
- You actually learn something, not just get an answer
ChatGPT has an “extended thinking” mode (looks like a clock icon). Use it for complex financial calculations - it forces the AI to reason step-by-step.
Tree of Thoughts (exploring multiple paths) #
This one’s wild. You have the AI explore multiple approaches, evaluate each one, then combine the best parts into one optimal strategy.
Here’s an example for debt payoff:
“You’re a financial counselor who specializes in debt elimination. I’ve got:
- $35,000 student loans at 5.5%
- $8,000 credit card debt at 18%
- $12,000 car loan at 4%
- $2,000/month surplus after minimums
Use Tree of Thoughts to find the best payoff strategy:
Step 1: Brainstorm three approaches:
- Branch A: “Debt Avalanche” - highest interest first (pure math)
- Branch B: “Debt Snowball” - smallest balance first (psychology)
- Branch C: “Hybrid” - balance transfer the credit card to 0%, then avalanche
Step 2: Evaluate each:
- Calculate total interest and timeline for each
- What are the psychological pros/cons?
- Risk factors (what if I lose my job?)
- Which builds momentum vs. saves most money?
Step 3: Combine the best parts into one “Golden Path” that balances:
- Math efficiency
- Psychological wins
- Risk management
Step 4: Give me the complete plan:
- Month-by-month payment schedule
- Total interest saved vs. just paying minimums
- Milestones to celebrate
- Backup plan if income changes
Show your reasoning for everything.”
Why this is powerful: You get the benefits of multiple strategies combined. Pure optimization + psychological momentum + creative alternatives. All customized to YOUR situation.
Build a prompt library (I’m using Obsidian for that) #
Here’s what nobody tells you: once you find a prompt that works great, SAVE IT. Build a collection.
Good ones to save #
Monthly budget review: “Act as my CFO. Review last month’s spending [paste data]. Compare to my budget [paste budget]. Identify top 3 areas where I overspent, explain why, give me specific fixes. Format: Problem | Why It Happened | Solution | Expected Savings.”
Investment rebalancing: “You’re a fee-only advisor. Target allocation: 70% stocks, 25% bonds, 5% cash. Current allocation: [paste data]. Calculate exactly what I need to buy/sell to rebalance. Give me a specific action list with dollar amounts.”
Tax optimization: “Act as a tax specialist for middle-income earners. Income: $X, current deductions: $Y, investments: $Z. Give me the top 5 ways to reduce my tax burden before year-end. Prioritize by savings amount and how easy they are to do.”
Emergency fund check: “You’re a risk management specialist. Monthly expenses: $X, current emergency savings: $Y, job stability: [level], dependents: [number]. Is my emergency fund adequate? If not, how much more and how fast should I build it?”
Where to keep them #
Keep it simple:
- Text file (old school works)
- Note app (Obsidian, Notion, Evernote)
- Fabric tool (has an
improve_promptfeature) - Spreadsheet (organized by topic)
Tag them so you can find stuff:
- By topic (budgeting, investing, debt, taxes)
- By complexity (beginner, advanced)
- By time needed
- By output format
Making your prompts better over time #
Your first attempts won’t be perfect. That’s normal. The key is getting better at it.
The cycle:
- Try a prompt on a real question
- Look at what you got back - was it useful?
- Figure out what was missing
- Adjust the prompt (more context, better formatting, whatever)
- Try the new version
People like Daniel Miessler, Joseph Thacker, and Eric Pope have built entire frameworks around this. Miessler’s Fabric tool has this improve_prompt feature that’ll analyze your prompts and suggest how to make them better.
Common fixes:
- Too vague? Add more context and numbers
- Too long? Break it into multiple prompts
- Wrong tone? Adjust the persona and output requirements
- Missing stuff? Add examples
- Getting bad answers? Use Chain of Thought so you can see where it’s going wrong
The real skill here is clarity #
At the end of the day, this whole prompting thing is really about thinking clearly. When you get good at articulating exactly what you need from an AI, you also get better at:
- Asking good questions in real life
- Making clearer financial decisions
- Talking to actual financial advisors
- Understanding your own goals
Better prompting = better thinking. It’s not just an AI hack - it’s a life skill.
Your action plan for 2026 #
Ready to actually use this stuff? Here’s what to do:
- Pick ONE financial question you’re stuck on (budget, investing, debt, whatever)
- Write a complete prompt with all four pieces (persona, context, output format, examples if needed)
- Run it in ChatGPT, Claude, or whatever AI you use
- Save both the prompt and the results
- Create 5 core prompts for questions you ask a lot
- Test and tweak each one
- Organize them by topic
- Try Chain of Thought on a complex calculation
- Test Tree of Thoughts on a big decision
- Build a complete library covering your whole financial life
- Share prompts with friends (help them level up too)
- Track if this actually improves your results
- Check out Fabric or other prompt tools
Look, AI is changing how we deal with money. But it only helps if you know how to use it. Most people are getting garbage advice because they’re asking garbage questions. And remember, it’s not just for dealing with money, it’s for everything else too.
You don’t have to be most people.
Start building your prompt library today. Your financial freedom literally depends on asking the right questions the right way.
What’s the first financial decision you’re gonna tackle with this?