Prompts for Financial Planning: Inputs, Scenarios, and Checks

Prompts for financial planning work best when they ask an AI to show its inputs, flag its assumptions, and stop short of decisions that belong to a licensed professional. A well-framed prompt turns a general-purpose model into a first-pass analyst: it lists what you gave it, builds a scenario table, and marks every estimated cell rather than blending assumptions into a clean number. Tools like fenbs let you save the prompts and reviewed assumptions that proved reliable, so every new session starts from the same tested baseline instead of rebuilding the groundwork.

The prompts below follow one rule: ask for the reasoning, not just the result. A model that shows how it arrived at a projection gives you something you can verify. One that returns a polished figure without the working gives you nothing to check against.

What AI can and cannot do in financial planning

AI assistants are reliable for organizing information, building scenario tables, explaining financial concepts in plain language, and spotting internal inconsistencies in numbers you provide. They cannot access your real accounts unless you connect a tool that can. They also have no knowledge of your actual tax situation, the current rate environment, or recent changes to US law — and they will produce confident-looking projections that are wrong if the inputs are wrong.

A model may misstate a tax rule, compound an interest rate incorrectly, or combine two unrelated figures because they appeared near each other in training data. Checking the working — not just the answer — is the only reliable defense against this. Ask the model to show every formula and label every assumption explicitly, and treat any figure that lacks a source as a prompt to verify.

Use AI for the analytical groundwork: listing assumptions, comparing two approaches as a table, explaining what a term means, flagging what you still need to find out. For anything that touches taxes, investments, or retirement under US law, a CPA, CFP, or RIA review is not optional and these prompts are not a substitute.

Prompts for gathering planning inputs

A plan that starts with incomplete inputs will have errors that are invisible until something goes wrong. The first step is surfacing what you actually know versus what you are assuming. These prompts ask the model to build the list of needed inputs and flag every gap before it touches a number.

I am preparing a personal financial plan. Help me build a complete list of inputs I need before modeling any scenario.

Organize the list into: monthly income sources, fixed expenses, variable expenses, outstanding debts (balance, interest rate, minimum payment), savings and investment accounts (current balance, monthly contribution), and major financial goals with a rough timeline.

Do not estimate any number I have not provided. Mark each item I have not yet given you as [needed].

Here are my planning inputs: [paste your numbers]

Before building any projections, list every assumption that is not covered by the numbers above. For each one, write the question that would settle it. Do not fill in a default value — wait for my answer.

Running these two prompts before any projection catches the gap between what you think you provided and what the model actually has. Save the settled input list as a Decision in your planning project so future sessions begin from the same agreed baseline, not from whatever you happen to paste in that day.

Prompts for comparing scenarios

Scenario comparison is where AI saves the most time in planning work. A side-by-side table of two approaches takes a few minutes to produce and hours to build by hand. The key is asking the model to label every cell with its formula so you can confirm that Scenario A and Scenario B are actually comparable.

Avoid asking for a recommendation. Ask instead for a comparison table that shows what each scenario costs, what it produces, and what conditions would make one better than the other. The recommendation stays with you and your advisor.

Using only the inputs I gave you, build a comparison table for these two scenarios:

Scenario A: [describe the first approach, e.g., pay off all credit card debt before saving anything else]
Scenario B: [describe the second approach, e.g., split extra cash evenly between debt payoff and emergency savings]

For each scenario show: monthly cash surplus or deficit, estimated months to reach the goal, and total interest paid. Label every cell with the formula or assumption behind it. Mark anything that depends on an estimated future figure as [estimated].

Take Scenario A from the table. Show me how the outcome changes under three conditions:
— my monthly income drops by 10%
— my monthly income rises by 10%
— an unexpected expense of $[amount] appears in month [n]

Present the results as a table with one row per condition. Explain the key difference from the base case in one sentence per row.

A sensitivity table is useful for finding fragile assumptions. If a single unexpected expense collapses Scenario A entirely, that is worth knowing before you commit to it. For similar structured comparison prompts applied to project decisions rather than finances, see prompts for project management AI.

Prompts for checking calculations

AI can review work a spreadsheet or a previous session already produced. Frame these as audits rather than requests: ask the model to walk through the logic step by step and flag anything that does not hold. An audit prompt that asks the model to commit to each step is harder for it to blur past an error than one that asks for a summary.

I want to check this calculation: [describe the calculation or paste the formula].

The inputs are: [list each input and its value].

Step through the math and tell me whether the result is correct. If you find an error, explain what went wrong and show the corrected result. If you are not certain about any step, say so rather than guessing.

Here is my financial plan summary: [paste summary or bullet points]

Check for internal inconsistencies — for example, expenses that add up to more than income, a savings goal whose timeline does not match the monthly contribution, or interest calculations that do not match the stated rate.

List each inconsistency you find, explain why it is a problem, and write the question that would resolve it. Do not suggest solutions — only surface the issues.

These prompts work best when you paste actual numbers rather than describing the situation in general terms. A specific number either checks out or it does not. A vague description gives the model room to assume, which defeats the purpose of the audit.

Prompts for capturing decisions and next steps

A planning session often ends with a short list of things to confirm, research, or do. If those next steps live only in the chat, they disappear when the session closes and are gone when you open a new conversation with a different model. Use this prompt at the end of every planning session to pull the decisions and open questions out before you stop.

Based on this planning session, please list:

1. Every assumption we settled on that I should record as a decision. For example: "I will target a six-month emergency fund before contributing to a brokerage account."
2. Every open question I still need to answer before the plan is complete.
3. Every action I need to take in the next 30 days, including the person or service to involve.

Use bullet points under each heading. Keep each item to one sentence.

Saving decisions outside the chat matters because they change the plan if they change. An assumption that lives only in a thread is invisible the next time you open a new session. Read system prompt examples for patterns that keep models consistent with saved rules across sessions, and prompt engineering for how to write instructions that hold when the conversation is long.

How to do this with fenbs

1. Create a free account and add a project for your financial planning work.
2. Paste each prompt that produced reliable output into a Skill in your project. Give it a short title and note what inputs it expects so it is ready to reuse.
3. Add a Decisions and rules entry for every planning assumption you confirmed: the rule, the date, and the reason. For example: "Emergency fund target is six months of fixed expenses before brokerage contributions — agreed after comparing Scenarios A and B, October 2026."
4. Connect your AI app over MCP from Integrations. Before each planning session, ask the assistant to read the project context so it sees your existing decisions before you paste in new numbers.
5. After each session, run the extract prompt above and record the new decisions and follow-up tasks in fenbs. The next session picks up from where the last one ended.

fenbs does not replace a spreadsheet or a financial advisor. It keeps the prompts that work and the assumptions you confirmed, so the analytical work builds on itself across sessions rather than starting from scratch each time. For broader work-task prompts that pair well with planning, see chatgpt prompts for work.