Watch on TikTok
Watch the Full Breakdown — @smobyday

Budgeting apps are fine. But most of them just track what you spent — they don't tell you what it means, what to cut, or whether you're actually trending toward your goals. ChatGPT can do all of that in a single conversation, with your own real data.

Here's the exact setup I use. Three prompts, one bank statement export, and you walk away with a "Wealth Glow-Up" report that's actually kind of entertaining to read. The prompts are engineered specifically to get useful, specific, non-judgmental output — not generic finance advice.

Watch →
Step 1 — Export Your Bank Statement as a CSV

Every major bank lets you download your transaction history as a CSV file. Here's where to find it:

Chase: Log in → Accounts → Download Account Activity → CSV → Select date range
Bank of America: Accounts → Download → Choose date range → CSV format
Wells Fargo: Accounts → Download Activity → CSV
Chime: Statements → Download → CSV

Pick the last 30–60 days so the analysis reflects your current habits, not some version of you from six months ago.

smobyday tip

No CSV option? Copy and paste your transaction list as plain text — ChatGPT reads it just as well. Takes 2 extra minutes but works great.

Step 2 — Upload to ChatGPT and Run the Initial Analysis

Open ChatGPT (the paid version handles file uploads directly; free version works if you paste the data as text). Upload your CSV, then paste Prompt 1 exactly as written — the specific wording is what drives the quality of the output.

Why prompt wording matters

AI tools respond to how you ask, not just what you ask. The prompts below are engineered with three specific techniques: role assignment (telling ChatGPT it's your money coach), explicit output format (numbered steps, tables, pie chart), and tone direction (conversational, not a lecture). Each of those choices changes the quality of what comes back.

Prompt 1 — The "Wealth Glow-Up" Report
I'm uploading my bank statement CSV for the past [30/60] days. Act as my personal money coach and create a "Wealth Glow-Up" spending report. Use a friendly, conversational tone — not a finance lecture. My financial goals are: [e.g., save $500/month, invest $200/month, pay off $2,000 debt by December, build a $5,000 emergency fund] Please do the following in order: 1. **Categorize every transaction** into: Housing, Food & Dining, Transportation, Entertainment, Shopping, Health & Wellness, Subscriptions, Financial (savings/investing), and Other. Show the total for each category. 2. **Create a "Spending Energy vs Goal Alignment" table** and label each category with one of the following: - "This amazingly aligns with your goals, baddie!!!" — intentional, goal-aligned spending - "Oooo spent a lot of coin here" — high spend worth reviewing - "Beige flag baddie" — neutral habit, neither helping nor hurting 3. **Generate a color-coded pie chart** using Python and matplotlib. Use a palette of warm, modern colors — purples, coral, gold, sage green, dusty pink. Add a legend, percentage labels on each slice, and a clean white background. Make it visually polished, not default style. 4. **Generate a horizontal bar chart** showing each spending category ranked from highest to lowest. Use the same color palette as the pie chart. Add dollar amounts as labels at the end of each bar. 5. **Give me 3 specific "glowup" moves** based on what you see in my actual data — not generic advice. 6. **Calculate how much I could redirect** toward my goals if I made those 3 changes. Be honest but encouraging. If something looks great, say so. If something needs a look, be direct but kind.
smobyday tip

ChatGPT Plus (paid) can generate real Python charts — actual rendered images, not descriptions. If you're on the free plan, it'll describe the chart instead. The paid plan is worth it just for this feature if you're serious about making this a monthly habit.

Step 3 — What the Output Actually Looks Like

Here's a sample of what comes back — the spending breakdown, the vibe labels, and the charts ChatGPT generates from your real data:

Sample Output — Wealth Glow-Up Report
Sample data for illustration — your real numbers will vary

The Glowup Moves are specific to your actual data. Not "cut back on coffee" but "you spent $312 on delivery apps — switching to meal prep 3x/week could save ~$200 and redirect it straight to your emergency fund."

Step 4 — If the Output Isn't Quite Right, Refine It

The first response isn't always perfect — and that's fine. AI prompting is iterative. If the output is too generic, too long, or missing something, these follow-up prompts pull it back in:

If it's too vague: "You gave me general advice — I need you to reference specific transactions from my data. Be more precise."

If the tone is too formal: "Make this more casual and direct. I want it to read like a text from a financially savvy friend, not a financial advisor."

If you want to go deeper on one category: Use Prompt 3 below.

If the glowup moves feel generic: "The glowup moves were too broad. Pull specific merchant names and dollar amounts from my statement and make the recommendations based on what I actually spent."

How prompt refinement works

Think of it as a conversation, not a form. Your first prompt sets the context. Your follow-up messages refine the output. The more specific you are about what's missing or what you want adjusted, the better ChatGPT corrects course — it remembers everything from the conversation so you never have to re-upload your data.

Step 5 — Go Deeper on Any Category (Bonus Prompt)

Once you have the Wealth Glow-Up report, you can drill into any category that stood out. This prompt is especially useful for dining, shopping, or subscriptions — the categories that tend to have the most hidden waste:

Prompt 3 — The Category Drill-Down
I want to look more closely at my [CATEGORY — e.g., Subscriptions / Food & Dining / Shopping] spending from this statement. Please: 1. List every transaction in this category with the merchant name, date, and amount 2. Group them by merchant — show me which ones I used most and the total I spent there 3. Flag any recurring charges that appear to be subscriptions I might have forgotten about 4. Tell me which specific merchants I could cut or reduce — and what I'd save per month 5. Give me one action I could take this week to reduce spending in this category Be specific. Use the actual merchant names and amounts from my data.
Step 6 — Do This Every Month

The real power of this system isn't a one-time report — it's running it monthly and comparing what changed. That's what Prompt 2 is for:

Prompt 2 — Monthly Momentum Check
Here is my bank statement for [MONTH]. Last month, my spending breakdown was: - [Category]: $[AMOUNT] - [Category]: $[AMOUNT] (paste last month's summary or upload the previous CSV) Last month my goals were: [e.g., cut delivery apps by $100, redirect $150 to savings] Please: 1. Compare this month to last month in a simple table — what went up, what went down, the dollar difference 2. Tell me directly: did I hit my goals from last month? 3. Flag any new "spending energy" concerns that didn't appear last month 4. Calculate my progress toward my bigger goals: [e.g., I want to save $5,000 by December — how am I tracking?] 5. Give me 2 new "glowup" moves for next month based on what you see Be conversational. Tell me what's working, not just what needs to improve.

This monthly cadence is what separates people who "tried budgeting once" from people who actually built new money habits. First month is insight. Second month is accountability. Third month is momentum — and by then you stop dreading looking at your account.

Why This Works Better Than Budgeting Apps

It uses your real data. Not estimates, not categories you manually tag — your actual transactions, analyzed automatically.

It's goal-aware. Mint and YNAB track what you spent. ChatGPT cross-references it against what you actually said you wanted to accomplish. That's a fundamentally different kind of feedback.

The prompts are engineered for tone. Nobody wants a spending report that feels like a tax audit. Specifying "friendly and conversational" in the prompt changes the entire register of the output — which means you're more likely to read it, absorb it, and act on it.

You can iterate. If the output misses the mark, you say so and it adjusts in the same conversation. No app can do that.

It's free. You already have access to ChatGPT. No $12/month budgeting app needed.

smobyday tip

Save your monthly reports as PDFs in a folder called "Money Check-Ins." Looking back at month 1 vs. month 4 is genuinely motivating once the numbers start moving.

Privacy Note

You're uploading real financial data, so a few things worth knowing:

→ ChatGPT (paid) doesn't train on your conversations by default. Verify in Settings → Data Controls → turn off "Improve the model for everyone."
→ To be extra cautious, anonymize the CSV by removing merchant names and keeping only amounts and categories — the analysis still works.
→ Never upload account numbers, routing numbers, or SSNs. The transaction list is all you need.