How to Make Money with AI: What Actually Works and What Does Not
Most 'make money with AI' advice describes a business that only worked for the person selling the advice. Here are the models that survive contact with reality.
You do not need another budgeting app. You need someone to look at three months of statements and tell you the truth, which is a conversation, not a dashboard.
Budgeting apps fail for a boring reason: they require ongoing maintenance from someone who did not enjoy the setup. Categories drift, transactions go unclassified, and eight weeks later the app is a monthly charge on the statement you are not reading.
A conversation works differently. You paste in what you actually spent, ask real questions about it, and get answers about your money rather than about money in general.
Export a CSV from your bank rather than sharing credentials — no assistant should ever be asked to log into a bank account. Then strip what is not needed for the analysis: full account numbers, sort codes, card numbers, and your address. Merchant name, date, and amount are enough to do all of this.
Here are three months of transactions (merchant, date, amount). Categorize them, and show me: monthly totals by category, which categories are stable versus volatile, my ten largest single expenses, and my ten most frequent merchants. Then tell me the three things that would most surprise me about my own spending.
Three months, not one. A single month is atypical by definition — everyone’s month has something unusual in it.
Now split my spending into fixed commitments, genuine variable spending, and things I probably think are fixed but aren't. That third category is the one I care about.
This is reliably where the money is, and it is the single highest-return hour in personal finance.
Find every recurring charge in these transactions, including annual ones and anything that renewed at a higher price than it started at. Give me the monthly and annual cost of each, flag any I'm paying for twice in different forms, and sort them by annual cost. Include anything under $5 — those are the ones people never notice.
For each of these, ask me one question that would tell you whether I should keep it. Then sort them into keep, downgrade, negotiate, and cancel — and write the three messages I need to send.
The messages matter. Most people know which subscription to cancel and do not cancel it, because writing the email is the friction.
Find every fee, charge, or penalty in these transactions — overdraft, ATM, foreign transaction, late payment, delivery, service charges, anything. Total them for the period, annualize it, and tell me which are structural (wrong account type, wrong card abroad) versus behavioral (timing, forgetting). Structural first — those are one-time fixes.
Structural fees are the best money in personal finance: fixed once, saved forever.
Identify my recurring small purchases — the ones that happen weekly or more. Show each as an annual figure. Don't tell me to stop buying coffee; tell me which of these I'd genuinely miss and which are pure autopilot, based on how the frequency changes when other things in my week change.
The annualized number is the point. Nobody changes a $4 habit; some people change a $1,000 one.
This is where a conversation beats a spreadsheet, because you can argue with it.
I'm deciding whether to move to an apartment that's $250 a month more but cuts my commute from 55 minutes each way to 15. Here's my budget. Work through the full financial picture including transport costs I'd save, then the non-financial side. Argue both cases properly, then tell me what additional information would actually change the answer.
Given my income and spending, model three savings rates — conservative, realistic, and uncomfortable. For each, show what it means month to month and what I'd be giving up. Be honest about which one someone with my actual spending pattern is likely to sustain.
File Assistant for working from a statement export, the Financial Analyst assistant for the scenario conversations, AI Research when you need current product terms rather than a recalled answer, and AI Goals for the savings target — because the plan is easy and month seven is not.
Share an export with identifying details removed, never credentials. Merchant, date, and amount are sufficient for every analysis here.
For analysis and decisions, it is better — you can ask follow-up questions. For automatic ongoing tracking, an app that connects to your accounts still wins. Many people are best served by an app for the data and a conversation for the thinking.
It will explain, compare, and model scenarios. Regulated advice — what to invest in, what to do about debt, how to handle tax — needs a qualified professional.
The full pass quarterly. The subscription audit is worth doing whenever you notice you have not thought about it in a while, which is usually a sign it is overdue.
Export three months, strip the identifiers, and ask what would surprise you. That single question has produced more useful answers than most budgeting software, and it takes about fifteen minutes.
General information, not financial advice.
Try it in ChatUp
Run the prompts above against the model that suits the task, keep the useful context across chats, and pick it back up on any device.
Try for Free