Your first AI wins: 5 everyday tasks

When you finish, you can use AI on five analyst tasks with strong prompts and a clear check for each.

Practice with public or sample data

Use public reports, sample datasets or made-up numbers while you learn. Do not paste confidential company figures, client positions or non-public information into a consumer tool. Use an approved tool only when your employer allows it for that data.

A strong analyst prompt includes the audience, the format, what to quote and what not to do. “Do not calculate” and “keep all numbers exactly as written” are protective phrases you will use often.

Wins one and two: filings and commentary

Weak: “Summarize this annual report.” Strong: “Summarize this public section for an analyst under these headings: business changes, risks, accounting policy changes, unusual items. Name the section for each point. Do not give an investment view.” What to check: every item against the section named, and whether anything important is missing.

For variance commentary, weak: “Explain these variances.” Strong: “Draft one-sentence commentary for each line above the threshold and list questions for the business owner. Do not invent causes.” Check that the commentary matches the table's sign and size. A favorable variance described as adverse is a classic slip.

Wins three and four: cleaning and formulas

Paste a small sample of messy data, not the entire file, and ask the tool to describe problems and propose cleaning steps in your software. Weak: “Clean this data.” Strong: “List the inconsistencies and give steps I can run in my spreadsheet. Do not change values.” Check by running the steps on a copy and comparing totals before and after.

To understand an inherited formula, ask for a step-by-step explanation and a list of ways it could fail. Check by testing it on a few rows where you already know the answer. If the AI's explanation and the test disagree, trust the test.

Win five: model review checklist

Describe your model's tabs and ask for a reviewer's checklist: hardcodes inside formulas, sign errors, unit mismatches, circular references and balance checks. You get a structure that is easy to run through, and you can adapt it for your team.

What to check: the checklist is only a prompt for your own testing. Run each check in the model. Document what you found. The AI did not review the model; you did.

Practice now

Download a public company's annual report section and run the summary prompt. Check three points against the text.
Make a small table of five lines with budget and actual figures, and ask for commentary plus owner questions. Verify signs and sizes.

Use any AI assistant you have. Do not paste private, client or patient information into it.

Check yourself

1. What is a safe way to test a data-cleaning plan?

2. Why add 'do not invent causes' to a variance prompt?

3. What does the AI model review checklist do?

Ask the AI tutor

Ask anything about this lesson, in any language. The tutor is an AI: it can be wrong, and it never needs your personal details.

Next lesson: Guardrails for financial analysts

AI-drafted. Checked by an automated validator and a separate AI reviewer; a person has not read every lesson. Translations and tutor answers are AI-made. Found a mistake? Tell us. AI can be wrong. General information only, not professional advice. Not affiliated with any AI company named. Current University is not accredited.