Check how much is left on a car loan with ChatGPT Work and Grok Bot, and give up at the lender's login (failure)
- Posted
- by John Helmuth (@johnhelmuth_)
- Last checked
- Method:
- How we verify
Evidence: Described by posterThe person describes what happened in words, without screenshots or video.
FailureThe source says the agent did not get the job done.
What went wrong: Neither agent got past the lender's sign-in; the poster checked on their phone in 30 seconds.
Disclosure: This is our product: realagentcases is run with Grok Bot. Included because the source is public and verifiable.
Outcome
John Helmuth asked both ChatGPT Work and Grok Bot how much time was left on their Ford Explorer loan with Ford Credit. Both first searched email for a statement, found none and went to the Ford Credit site, then got stuck at the login. ChatGPT's secure form didn't work and its cloud browser never showed a keyboard; Grok Bot handed over its cloud browser with a working keyboard, but after Helmuth entered their email and password it stayed stuck on the sign-in screen. After 5 minutes on both, Helmuth logged in on their phone's browser with a passkey and saw the loan terms in about 30 seconds.
Agent
Prompt Recreated prompt
The source didn't share its exact prompt, so we wrote this from what it describes.
Run this task now, exactly as written below. Everything you need is here, so start straight away and only stop if something required is missing or a rule below tells you to. Goal: Tell me how much is left on my loan and when it ends, using my statements or the lender's site. Fill in before running: - Lender and what the loan is for: [DETAILS] - Where my statements might be (email, files): [DETAILS] Rules: - Look for a recent statement in my email or files first. - If you need the lender's site, tell me and let me sign in myself; if the sign-in doesn't work after one try, stop and tell me. - Give me the balance, payments left and end date, and the date of the figures. - Don't change any account settings or make payments. - If a site asks me to sign in, pause and let me do it myself (including any CAPTCHA or two-step code); never guess or store passwords. - Check facts that can change (prices, opening hours, dates, availability, rules) against the official source, and tell me anything you couldn't verify. Output: Balance, payments remaining and payoff date, with the source. You'll know it worked when: I know where my loan stands in less time than checking myself.
Customize this prompt
Answer these questions and we'll put your answers into the prompt for you. You can leave any of them empty.
Your prompt
Run this task now, exactly as written below. Everything you need is here, so start straight away and only stop if something required is missing or a rule below tells you to. Goal: Tell me how much is left on my loan and when it ends, using my statements or the lender's site. Fill in before running: - Lender and what the loan is for: [DETAILS] - Where my statements might be (email, files): [DETAILS] Rules: - Look for a recent statement in my email or files first. - If you need the lender's site, tell me and let me sign in myself; if the sign-in doesn't work after one try, stop and tell me. - Give me the balance, payments left and end date, and the date of the figures. - Don't change any account settings or make payments. - If a site asks me to sign in, pause and let me do it myself (including any CAPTCHA or two-step code); never guess or store passwords. - Check facts that can change (prices, opening hours, dates, availability, rules) against the official source, and tell me anything you couldn't verify. Output: Balance, payments remaining and payoff date, with the source. You'll know it worked when: I know where my loan stands in less time than checking myself.
Scroll to see the whole prompt. Your answers show in green.
2 blanks are still empty: Lender and what the loan is for; Where my statements might be (email, files). They'll stay in brackets in your prompt, so the chat tool may ask you about them.
What you type stays in this browser tab. We don't send it or save it, and it's gone when you leave the page. If you use a "Try this in" button, your answers go to that tool as part of the prompt.
Try this in:
This prompt has blanks to fill in, like [DETAILS]. We'll copy it instead of filling it in, so you can edit it before you send.
Opens in a new tab. We don't see what you send.
These links are only a shortcut. We're not affiliated with OpenAI, Google or xAI, and your results may not match the source.
Optional stronger version
Our rewrite with extra checks. The poster didn't test it.
Run this task now, exactly as written below. Everything you need is here, so start straight away and only stop if something required is missing or a rule below tells you to. Goal: Tell me how much is left on my loan and when it ends, using my statements or the lender's site. Fill in before running: - Lender and what the loan is for: [DETAILS] - Where my statements might be (email, files): [DETAILS] Rules: - Look for a recent statement in my email or files first. - If you need the lender's site, tell me and let me sign in myself; if the sign-in doesn't work after one try, stop and tell me. - Give me the balance, payments left and end date, and the date of the figures. - Don't change any account settings or make payments. - If a site asks me to sign in, pause and let me do it myself (including any CAPTCHA or two-step code); never guess or store passwords. - Check facts that can change (prices, opening hours, dates, availability, rules) against the official source, and tell me anything you couldn't verify. - If you can't get in, tell me the quickest way to find it myself, such as the lender's app. Output: Balance, payments remaining and payoff date, with the source. You'll know it worked when: I know where my loan stands in less time than checking myself. Before you finish: re-read my goal and constraints, check your answer against each one, and list anything you assumed or couldn't confirm.
Source
https://x.com/johnhelmuth_/status/2096596752790954437
Related cases
- Catch a 2-litre engine-oil overcharge on a car service bill with ChatGPTRecreated promptEvidence: Described by posterBeginner-friendly
Billed for a 6-litre pack of engine oil after a service on a Mahindra XUV 3XO at an authorised dealership, Gurpreet Singh Sodhi photographed the invoice and uploaded it to ChatGPT. It helped explain the invoice, identify the exact engine and its documentation, and prepare questions for the dealership. The dealership investigated, admitted the error, apologised and is re-issuing the invoice for 4 litres; the corrected bill is almost ₹1,500 lower.
ChatGPT · Personal · Source date:
- Catch a returned $2,350 insurance reimbursement check buried in 80k unread emails with Grok BotRecreated promptEvidence: Described by poster
Within five minutes of setting up Grok Bot, a poster with about 80k unread emails had it flag one message from four days earlier: the insurance company said their windshield replacement reimbursement check had been returned by the post office and needed a new address. The poster says it was the only email the bot flagged and that it saved them $2350.
Grok Bot · Personal · Source date:
- Estimate the year's federal, state and local tax still due from a folder of tax documents with Claude Cowork, before the accountant reportsEvidence: Described by poster
Anthony Garcia's taxes had got complicated enough that Garcia hired an accountant, who was still a month and a half away from giving the full picture. Garcia copied all the tax files, some electronic PDFs and some scans of paper copies, into one folder, pointed Claude Cowork at it and asked for an estimate of the 2025 taxes due. Claude read all of them. In 30 minutes it produced a summary table of tax liability, amount withheld and balance due, with a thorough breakdown that referred to boxes on the tax forms.
Claude Cowork · Personal · Source date: