AI and MCP

Month-end spending review prompts for AI assistants

Five prompts for the last day of the month: what changed, what repeated, what surprised you, what to fix next, and a one-line summary.

Editorial illustration for Month-end spending review prompts for AI assistants

A month-end spending review with an AI assistant takes five prompts on one pasted table: what changed, what repeated, what surprised you, what to fix next month, and a one-line summary. Paste your category totals for this month and last month, run the prompts in order, and write down one action. Ten minutes on the last day of the month is plenty.

Paste this table first

Use your own numbers. This hypothetical table is what the prompts below work on.

CategoryThis month (₹)Last month (₹)Entries
Rent18,00018,0001
Groceries9,2008,40014
Eating out4,5003,1009
Transport2,8003,00022
Subscriptions1,2971,2974
Shopping6,0002,2005

The totals are ₹41,797 this month and ₹35,997 last month, a rise of ₹5,800. Ask the assistant to confirm both before you go on.

Prompts 1 and 2: what changed and what repeated

Prompt 1: "Compare the two month columns. List each category whose change is above ₹500, largest first, with the rupee change. Ignore smaller moves. Show the subtraction for each."

In the sample, it should return Shopping +₹3,800, Eating out +₹1,400 and Groceries +₹800. Transport (−₹200) falls below the line.

Prompt 2: "From this list of my ten most frequent merchants this month, tell me which ones also appeared last month and how many times each. Separate fixed monthly payments from habits." This needs a merchant list, which your statement or tracker can give you.

What surprised you, and what should you fix?

Prompt 3: "Ask me up to three short questions about the biggest changes before you comment on them." Answer honestly. Shopping may have been a festival outfit, not a habit.

Keep your answers short and factual: "festival clothes for two, one-off" or "ordered in on three late nights". The assistant uses those notes to separate a habit from an event, and the same notes jog your memory a month later.

Prompt 4: "Based on my answers, suggest one change for next month in a single category, with a rupee target and the arithmetic. Do not suggest cutting rent or other fixed bills. Offer only one change."

A good reply looks like this: cap Shopping at ₹3,500 next month, which is ₹2,500 below this month's ₹6,000 and still ₹1,300 above last month's ₹2,200. Check it: 6,000 minus 3,500 is 2,500, and 3,500 minus 2,200 is 1,300. If the target sits below what you spend in an ordinary month, it is a wish and not a plan, so ask the assistant to make it realistic.

Prompt 5: the one-line summary

Prompt: "Write one sentence I can paste into my notes: the total, the biggest rise, and my single action for next month." Expect something like: Spent ₹41,797, up ₹5,800, mostly shopping; next month shopping target ₹3,500.

Paste that sentence into the same note every month. After three months you have a short diary of totals, rises and actions, which tells you more than any single review can.

What if my categories are messy?

Fix them first. If half of Shopping is really Groceries, the prompts will confidently discuss the wrong problem. Ask the assistant to list entries whose merchant name does not fit their category, correct them in your tracker, then paste the table again.

A last-day routine

  1. Export or copy category totals for both months.
  2. Check the grand totals against your statements.
  3. Run prompts 1 to 5 in the same chat so the context carries over.
  4. Save the one-line summary and the single action.
  5. Look at that action again on the 8th of next month, not on the 30th.

If you use SpendixAI, the 25 ready-made reports and the Report builder give you the category totals and the two-month view without retyping. If you connect Claude or ChatGPT through MCP, you choose the scope and can start read-only.

SPENDIXAI JOURNAL

Useful ideas start with a clearer picture of your own spending.

Explore your reports