Which tool to choose
You don't need to install anything special. Choose based on how you work.
You want a tailored plan, free and right away, starting from your numbers → a general conversational assistant (ChatGPT, Claude or Gemini). It's the right option for 90% of cases: you paste your data by hand, no bank connection, no cost. The free version of ChatGPT handles this task well. It has one limit: after a certain number of closely-spaced messages on the most advanced model you get moved to a simpler one, but for building a savings plan you won't notice.
You want the app to read your account transactions on its own and categorize them → a budgeting app with built-in AI. Here, though, there's a concrete obstacle: ChatGPT's native finance feature hasn't arrived in Italy yet. The Personal Finance feature launched as a preview only in the United States and reserved for paying subscribers, relying on a service (Plaid) that connects thousands of American financial institutions. Translation: in Italy, today, you can't use bank automation inside ChatGPT.
You're a beginner and the idea of a spreadsheet blocks you → an app like Cleo, which talks via chat. Cleo uses a chatbot to show your spending analysis as a conversation, more digestible than a classic budget tracker. Here too a geography limit: it's designed for the United States and the United Kingdom, so in Italy the account connection may not work.
For readers in Italy the practical advice is clear: start with the free conversational assistant. It's available, it's free, and it does 80% of the real work, namely building your plan and keeping you on track.
How to do it
From browser or app, the path doesn't change.
- Gather three pieces of data: net monthly income, fixed expenses (rent/mortgage, utilities, subscriptions, installments), goal (amount + date by which you need it).
- Open ChatGPT, Claude or Gemini. Start a new conversation.
- Paste the prompt below, replacing the figures with your real ones.
- Read the plan. The AI will almost always propose a proven method: the 50/30/20, that is 50% of income to needs, 30% to personal spending and 20% to savings, adapted to your goal.
- Ask for the adjustments: "make the plan more aggressive", "I can't touch the gym", "I have an unexpected €300 expense this month".
The operational syntax:
Act as a personal finance analyst. I need to set aside €4,000
within 10 months to buy a used car.
Here's my monthly situation in euros:
- Net income: 1,800
- Rent + utilities: 750
- Groceries: 350
- Transport: 120
- Subscriptions (streaming, gym, phone): 95
- Free time and restaurants: 250
- Other/unexpected: 100
Calculate whether the goal is realistic within the 10 months. If it isn't, tell me by
how much to push the date or which expenses to cut. Propose a month-by-month plan
with the exact amount to set aside, indicating where to recover it.
Keep the plan sustainable: I prefer gradual cuts to total
sacrifices. Use only the numbers I gave you.
Check after the plan: make sure the amount to set aside each month, multiplied by the months, actually reaches the target figure. If the numbers don't add up, the AI got the arithmetic wrong (it happens): have it redo it by asking "verify the total sum of the plan".
A useful addition: paste the last two months' statement and ask it to track down waste. The AI analyzes the transactions and flags forgotten subscriptions or recurring expenses that have increased.
A concrete example
Marta, 28, wants €1,200 for a trip in July. It's February: five months. Net salary €1,500, fixed expenses €980, free time about €300, the rest a cushion.
She pastes the data into ChatGPT free. The AI tells her right away that €240 a month is needed and that with €220 of real margin she nearly gets there, but it's "tight". Instead of imposing drastic sacrifices, it proposes a gradual cut. This reflects a principle these tools often apply: too rigid a plan tends to fail, because people keep up small realistic habits better than big sacrifices.
Marta admits she orders food delivery three times a week. The answer isn't "stop": it's going from three orders to one. That cut frees up about €80 a month. ChatGPT builds her the plan: €200 set aside the first two months, €250 in the last three, with an automatic transfer to a separate account the day after payday. By the end of June Marta has €1,180. The last €20 she covers by skipping two aperitifs. Trip paid for, without having felt punished for five months.
When it does NOT work (and how to fix it)
If the AI makes up the numbers
Models get calculations wrong more than people think. Don't trust the total at a glance: ask "show me the sum month by month and the final total" and recheck with your phone's calculator. If the plan doesn't close at the right figure, have it redo it specifying the error.
If it gives you generic textbook advice
This happens when the data you gave is vague. The detail makes the difference: without numbers ChatGPT suggests a general plan, with your real numbers it analyzes the spending categories, spots the recurring items and estimates concrete savings margins. Paste precise figures, not "I spend little on going out".
If you want it to read the account on its own but you're in Italy
ChatGPT's bank connection isn't available here, and in any case it's read-only: it reads balances and transactions, doesn't see full account numbers and makes no changes. The workaround that works today: export the CSV or PDF of transactions from your banking app and paste it into the chat. You get the same analysis, done by hand.
If you can't stick to the plan
The problem isn't the plan, it's the missing automation. Set up a recurring transfer to a deposit account or a separate space the day after your salary lands: the money disappears from view before you spend it. The AI tells you how much, the bank executes it.
A tip from someone who actually uses it
Don't ask the AI "how do I save": it'll give you the same waffle you find everywhere. Ask it to do the math on your numbers and to tell you the uncomfortable truth: "is this goal in the timeframe I want realistic or am I fooling myself?". The real value isn't the motivational advice, it's the cold calculation that tells you whether 5 months or 8 are needed. And update the plan every month by pasting your real expenses: a static plan dies, one that recalibrates carries you to the finish line.
Frequently asked questions
Do I have to pay for a ChatGPT subscription to make a savings plan?
No. The free version is enough to build and update the plan. ChatGPT calculates a savings rate on your income and outflows and suggests different levels (conservative, moderate or aggressive) even without a subscription. The only free limit is the number of closely-spaced messages, which for this use won't affect you.
Which assistant is best among ChatGPT, Claude and Gemini for money?
For building a plan and doing the math they all work fine: choose the one you already use. ChatGPT is the most widespread and documented for personal finance; Claude tends to be orderly in step-by-step reasoning; Gemini is handy if you live inside the Google ecosystem. The practical difference is minimal compared with the quality of the data you put in.
Is there an app that saves for me automatically?
Yes, but watch out for geography. Tools like Cleo automate set-asides via chat: the free plan covers spending analysis, budget monitoring and savings goals, the advanced features start from a few dollars a month. In Italy the connection to local accounts is their weak point. For real automation, today it's best to set up a recurring transfer at your bank: same effect, zero cost.
Can I trust AI for important decisions about my money?
For organizing savings, yes. But don't confuse it with an advisor: OpenAI itself writes that the finance feature doesn't replace professional advice. Use it to do the math and keep yourself disciplined; for investments, mortgages or significant tax choices, check with a qualified person. The real risk isn't that the AI steals your data, it's that you act on a wrong calculation of its without rechecking it. That check is free and takes two minutes: always do it.