20 questions to ask AI about your personal finances (with real answers)
Most people use AI for everything except the one area where it could change their life the most: their money. Not because they don't want to — but because they don't know what to ask.
This is a practical list of 20 questions that become genuinely powerful once your bank is connected to Claude or ChatGPT through OpenBudget. These aren't generic prompts. Each one returns a real answer based on your actual transactions — with numbers, trends, and insights you'd never find by staring at a bank statement.
Why these questions hit different when AI has your real data# permalink to this section
Ask ChatGPT "what's my biggest unnecessary expense?" without bank access and you get a lecture about cutting coffee. Ask the same question with your real transactions connected and you get: "Adobe Creative Cloud at $54.99/month — you haven't logged in for 73 days."
That's the difference. Generic AI gives advice. AI with your data gives answers.
All 20 questions below assume your bank is connected to Claude or ChatGPT via OpenBudget's MCP integration. Setup takes about 5 minutes at openbudget.sh.
Category 1 — Spending awareness (questions 1–5)# permalink to this section
These are the "where did my money go?" questions. The ones you ask when you look at your bank balance and feel vaguely confused about how it got that low.
Q1 — "How much did I spend last month, broken down by category?"# permalink to this section
The first question everyone should ask. Before you can change anything, you need to see the full picture — not just your checking account balance, but all accounts, all cards, all categories.
What Claude returns: A complete breakdown across every connected account with month-over-month comparisons, share of total spend, and key insights about what changed.
Q2 — "What were my top 5 biggest expenses this month?"# permalink to this section
This question surfaces the individual transactions that drove your spending — not the categories, but the specific charges. In this example, the top 5 account for 57% of total monthly spend. Two of them are worth questioning: the Apple Store purchase ($399) is flagged as unusual because there's been no tech purchase over $100 in the previous 4 months, and the Nobu dinner ($247) is one of several high-end dining charges this month. Claude doesn't just list them — it contextualizes each one.
Q3 — "How much do I spend on restaurants vs groceries each month?"# permalink to this section
Food is almost always the most controllable category — and this question shows exactly how the balance has shifted. In this example, in January the user spent $1.79 on groceries for every $1 at restaurants. By May that ratio had flipped to $0.77. Claude spots the trajectory and projects that restaurants will permanently overtake groceries by June if the trend continues. That's not a number you'd ever calculate yourself.
Q4 — "Give me a full financial summary for April"# permalink to this section
One question, one complete monthly report. Claude returns total spend, income, savings rate, category breakdown with transaction counts, biggest individual transactions, daily average, busiest and quietest spending days, and a month scorecard with letter grades. The insight buried at the bottom is the kind of thing no dashboard would surface: "Without the Nobu dinner and 9 DoorDash orders, April would have looked exactly like March." That's actionable.
Q5 — "Do I spend more on weekends?"# permalink to this section
Most people have no idea how much their weekend behavior costs them. In this example, weekend days cost 65% more than weekdays — and the gap is widening month over month (61% pricier in January, 75% by April). Dining alone accounts for $180 of the monthly weekend premium. The actionable number: bringing weekend spending to just 1.3× weekday rate instead of 1.65× saves $140/month — $1,680/year.
Category 2 — Subscriptions and recurring costs (questions 6–8)# permalink to this section
Subscriptions are the spending category most people are most wrong about. Ask someone how much they spend on subscriptions and they'll say "$50 or so." Then you look at their bank.
Q6 — "List all my subscriptions and their monthly cost"# permalink to this section
The question that always surprises people. In this example, 11 active subscriptions total $231/month ($2,772/year) — and three of them show no usage in 60+ days. Adobe Creative Cloud ($54.99/month) costs more than Netflix and Spotify combined. Canceling just the three unused services saves $84/month, $1,007/year. Claude also flags a PlayStation Plus annual renewal coming in August — worth reviewing before it auto-charges.
Q7 — "Find recurring charges I might have forgotten about"# permalink to this section
What Claude does: Scans 5+ months of transaction history for charges that repeat on a regular cycle and cross-references them against usage signals. Groups results into "likely forgotten — cancel these" and "worth reviewing — possible overlap or overkill."
Example answer: "I found 6 recurring charges that look suspicious. LinkedIn Premium at $39.99/month has no detected profile views or job searches in 4 months — that's $480/year for a service you're not using. You're also paying for both iCloud 2TB and Dropbox Plus — two cloud storage services simultaneously. One is almost certainly redundant."
Q8 — "Which subscriptions have I not used in the last 3 months?"# permalink to this section
What Claude does: Looks for recurring charges where related in-app purchases, associated activity, or secondary signals suggest the service isn't being used. Returns a ranked list with estimated annual cost if you keep each one.
Why this question matters: Most subscription audits are manual — you scan a list and try to remember if you used something. Claude does it from your transaction history, which is more accurate than your memory.
Category 3 — Trends and patterns (questions 9–12)# permalink to this section
These questions find the patterns in your spending that are invisible in any single month — the slow creep, the weekly rhythms, the categories quietly growing in the background.
Q9 — "Is my spending increasing or decreasing over the last 6 months?"# permalink to this section
What Claude does: Pulls monthly totals for the last 6 months, calculates growth rate, identifies when acceleration started and what's driving it, and projects forward.
Example answer: "Increasing — your spending has grown $631/month since December, a 20% rise in 6 months. The trend accelerated sharply in March (+$245 that month alone). Dining is the single biggest driver — up $317/month vs December. At the current rate of growth, you'll be spending ~$4,200/month by October — a $1,000 increase from where you started the year."
Q10 — "What day of the week do I spend the most money?"# permalink to this section
What Claude does: Groups all transactions by day of week across your full transaction history, calculates average spend per day, and identifies which categories drive the peaks.
Example answer: "Saturday is your most expensive day — averaging $112/day vs your overall daily average of $67. Saturday dining alone is $287/month, driven by dinner with groups. Sunday is second at $88/day, mostly groceries and brunch. Monday is your cheapest day at $31."
Q11 — "Which category grew the most compared to last month?"# permalink to this section
What Claude does: Compares current month-to-date vs the same period last month for every category, ranks by absolute dollar change, and flags whether the growth is likely a one-off or a trend.
Example answer: "Travel grew the most — up $210 vs March, driven by the Delta Airlines flight on May 13. But that's likely a one-off. The more concerning growth is dining, up $155 vs last month — and unlike travel, this is the 4th month in a row it's increased."
Q12 — "How much have I spent on food this year?"# permalink to this section
What Claude does: Combines dining, groceries, food delivery, and coffee across all connected accounts for the year to date, shows month-by-month trend, and projects the full-year total.
Example answer: "You've spent $4,840 on food in January–May — $968/month average. That's on pace for $11,600 by December. Broken down: groceries $2,580 (53%), dining $1,810 (37%), delivery $450 (9%). Your lowest food month was February ($813). If every month looked like February, you'd spend ~$9,756 for the year — saving about $1,844 vs your current pace."
Category 4 — Savings and goals (questions 13–16)# permalink to this section
These questions connect your current spending to your future money. What are you actually saving? When will you hit your goal? What's the fastest path there?
Q13 — "How long will it take me to save $5,000 at my current rate?"# permalink to this section
The question that makes goals concrete. Claude doesn't just divide your surplus by the goal — it models three scenarios based on your actual spending trends. At current rate: 5 months (October). If dining keeps rising: 7–8 months (February). If you cut $300/month from easily identified waste: 3.6 months (September). The two most impactful actions are identified at the bottom: canceling Adobe and Audible ($70/month, two clicks) shaves 3 weeks off the timeline.
Q14 — "What's my average monthly income vs expenses?"# permalink to this section
What Claude does: Calculates average income and expenses over the last 3–6 months, shows the trend in your surplus over time, and flags whether that surplus is growing or shrinking.
Example answer: "Over the last 5 months, your average income is $4,800/month and average expenses are $3,743/month — leaving a $1,057 surplus. But the surplus has dropped from $1,582 in January to $952 in May — a $630 decline. If the trend continues, you'll be saving less than $500/month by September. The culprit is dining, up $337/month since January."
Q15 — "I want to save $2,000 in 3 months. Based on my spending, is that realistic?"# permalink to this section
What Claude does: Calculates whether your current surplus covers the goal, identifies the specific cuts needed if not, and gives you an honest assessment with a buffer.
Example answer: "Yes — but it requires deliberate cuts. You need $667/month. Your current surplus is ~$1,057/month, so the math works — as long as that surplus isn't quietly being spent. The real question is whether the $1,057 is actually being saved or drifting into small purchases. Dining is your biggest risk — it's grown from $295 to $632/month in 4 months. If that trend continues, your surplus shrinks fast."
Q16 — "What could I cut to save an extra $300/month?"# permalink to this section
Specific, ranked, and based on your actual patterns — not generic advice. In this example, Claude identified 5 cuts totaling $316/month. Crucially, it flags that cuts 2, 4, and 5 require zero lifestyle change — just canceling services you're barely using ($106/month for 3 clicks). The dining insight is the standout: you spent $295 in January and $632 in April — getting back to $400/month saves more than everything else combined.
Category 5 — Smart one-off questions (questions 17–20)# permalink to this section
These are the questions you only need to ask occasionally — but when you need them, no spreadsheet can answer them.
Q17 — "Am I on track to stay under $3,500 this month?"# permalink to this section
What Claude does: Takes your current month-to-date spending, calculates your daily pace, compares it to what you'd need to hit your target, and gives you a concrete answer with the specific actions that would change the outcome.
Example answer: "You're 18 days in and have spent $2,104 — 60% of your $3,500 budget. At your current pace of $117/day you'd end the month at ~$3,510, just barely over. One fewer dinner out (~$80) keeps you under. Watch out: your Amazon Prime renewal on May 28 ($139) isn't in these numbers yet."
Q18 — "Compare my spending in April vs March — where did I overspend?"# permalink to this section
What Claude does: Side-by-side comparison of every category across two months, flags the biggest increases, and identifies the specific transactions driving them.
Example answer: "You spent $312 more in April. The culprits: one Delta flight (+$210), one Nobu dinner (+$187), and 4 extra Uber Eats orders (+$94). Those three things account for $491 — more than the total overage — because groceries and shopping were actually lower. Excluding the one-off travel, your core spending was $78 lower than March."
Q19 — "How much did I spend on Amazon this year?"# permalink to this section
What Claude does: Finds every Amazon transaction across all connected accounts, totals them by month, identifies any trends, and contextualizes the number against your total spend.
Example answer: "You've spent $1,247 on Amazon in 2025 — $249/month average, across 34 orders. January was your biggest Amazon month ($387 — likely post-holiday). The average order is $37. Amazon is currently your 6th largest merchant by spend, behind rent, restaurants, groceries, Equinox, and Delta."
Q20 — "If I keep spending at this pace, what will my total be by end of month?"# permalink to this section
What Claude does: Takes your current pace, projects forward through the end of the month, and flags any known upcoming charges (subscriptions, rent) that haven't hit yet.
Example answer: "At your current daily pace of $117, you'll end May at approximately $3,627. Add your Amazon Prime renewal ($139) and a Equinox charge due May 29 ($39) — both not yet posted — and you're looking at ~$3,805 total. That's $305 above your usual baseline. The main driver is dining — you've had 14 restaurant charges this month vs your usual 8."
How to ask these questions today# permalink to this section
All 20 questions require your bank data to be live in Claude or ChatGPT. Here's how to set that up:
Step 1. Create an account at openbudget.sh.
Step 2. Connect your bank via Plaid — takes about 2 minutes, supports 10,000+ banks.
Step 3. Install the OpenBudget MCP server for Claude or ChatGPT. Copy-paste commands, no technical knowledge needed.
Step 4. Start with question 1 or question 4 — the category breakdown or full monthly summary. Once you see what Claude can do with your real data, the other 19 questions will make themselves obvious.
What makes these questions work# permalink to this section
Notice what all 20 questions have in common: they're specific, they're comparative, and they're about patterns — not just totals. "How much did I spend?" is less useful than "How much did I spend vs last month and where did it change?"
The best follow-up question after any of these is usually: "What should I do about it?" — because Claude won't just surface the insight, it'll help you act on it.
OpenBudget connects your bank accounts to Claude and ChatGPT via MCP — so every question on this list returns a real answer based on your actual transactions. Get started at openbudget.sh →
OpenBudget
