ChatGPT agent builds a classroom valuation sheet
Warr reports ChatGPT agent revising a hardcoded Apple valuation sheet into a formula-based classroom model with source URLs after he requested corrections.
Original prompt
Shared by the source author. See source notes for context.
Using data from https://finance.yahoo.com/quote/AAPL/ and other sources, create a spreadsheet for me that does a valuation of apple's stock
Reach people exploring AI agents
See placements →What happened
Warr reports ChatGPT agent revising a hardcoded Apple valuation sheet into a formula-based classroom model with source URLs after he requested corrections.
Reported by Richard WarrHow the agent did it
- Specify three valuation models and a simple one-period teaching scope.
- Ask the agent to gather financial inputs and create a spreadsheet.
- Inspect the first result for formulas rather than hardcoded outputs.
Where a person stepped in
Warr detects hardcoded outputs, requests formulas and sources, compares results with his own analysis and treats the workbook as classroom material rather than investment advice.
Richard Warr tested ChatGPT agent while preparing an undergraduate lesson on equity valuation. He wanted a simple one-period spreadsheet demonstrating dividend discount, free cash flow to equity and free cash flow to the firm models. The purpose was teaching the approach, not producing a definitive investment valuation.
The first output contained hardcoded figures without formulas. Warr asked the agent to redo the work with formulas and URLs for the inputs. It then gathered information from sources including Federal Reserve rates, Damodaran’s equity-risk-premium material, cash-flow statements and Apple investor relations. Warr linked the resulting spreadsheet and also reported that the agent produced the LaTeX equations shown in his post.
The useful checkpoint is to inspect how numbers are derived before judging whether the final spreadsheet looks plausible. Source URLs and live formulas make assumptions and relationships easier to check or change. Warr compared the output with his own analysis and considered it usable for classroom assignments, while saying it was not sufficient grounds for investment decisions. The entry therefore records an educational spreadsheet workflow, with a correction round and human interpretation of growth and discount-rate assumptions; it does not endorse the historical values or predict the stock’s future price.
Prompt
Original prompt
Source: Blog · Richard Warr
Frequently asked questions
Why was the first output rejected?
It was hardcoded and lacked formulas, so the instructor could not inspect the calculation structure.
What did the revision add?
Formula-based calculations and URLs showing where the financial inputs came from.
Was the workbook used as an investment recommendation?
No. Warr frames it as a teaching example and says it is not adequate as the basis for investment decisions.
More details
Remaining steps
- Request a revision containing formulas and source URLs.
- Compare the workbook with the instructor’s analysis before using it in class.
About this account
The agent gathers current data from several financial sources, builds a spreadsheet and revises its modelling output after the instructor identifies a transparency problem.
Firsthand account by Richard Warr; original publication date verified against Primary page publication date. Outcomes are the author’s reports, not independently reproduced. Original prompt excerpt transcribed from the primary account; excerpt only, not the complete prompt. Preserve the source-specific context when adapting it.
Credit: Richard Warr
Source published 12 Sept 2025
Source checked 14 Oct 2026
Tools and topics
ChatGPT agent · Spreadsheet · Federal Reserve data · Apple investor relations
Teaching · Spreadsheet formulas · Source transparency
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.