ChatGPT agent shortlists Black Friday TVs
Alex Hughes used ChatGPT to clarify TV requirements, then agent mode to search retailers and deal guides, returning five matching choices and two cheaper alternatives without a purchase.
tomsguide.com
Original prompt
Shared by the source author. See source notes for context.
Using my requirements set out above, find the best Black Friday deals on TVs.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Alex Hughes
- Original account
- Read the source ↗
- Published
- 12 Nov 2025
- Source link checked
- 15 Oct 2026
- Reported result
- Hughes reports seven suitable candidate offers and a final recommendation after roughly twenty minutes; no purchase is documented.
- Human involvement
- Hughes answered preference questions and reviewed the shortlist; no checkout or independent feature audit is reported.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Clarify TV preferences in regular Thinking mode before browsing.
- Switch to agent mode and search retailer offers against those requirements.
- Review five main choices and two cheaper alternatives, then request a final recommendation.
Alex Hughes wanted a television within a £300 budget but did not know which specifications mattered. He began in ChatGPT Thinking mode, answered nine preference questions and received an initial shortlist. That conversational preparation established the requirements; it was not the agent run itself.
Hughes then switched the same conversation to agent mode and asked it to find Black Friday deals against those requirements. The agent searched deal guides and major retailers and returned five options, plus two below £250 that met as many preferences as possible. Only one of the main choices repeated the earlier conversational suggestions.
The results included purchase links, explanations of suitability and the claimed discount. Hughes asked for a final recommendation and reports that the whole process took about twenty minutes, mostly spent waiting for research while he did other work.
No television purchase or delivered product is documented. The account does not independently validate every feature, reference price or discount. Its concrete outcome is a researched shortlist under explicit preferences, showing how clarification can precede delegated browsing. Historical offers in this test should not be treated as current buying advice.
Prompt
Original prompt
Source: Blog · Alex Hughes
Frequently asked questions
Which part used agent mode?
The retailer and deal-guide research followed an earlier Thinking-mode conversation that clarified requirements.
How many alternatives were returned?
Five main candidates plus two options below £250, according to the author.
Did he buy a TV in the account?
No completed purchase or delivered product is established.
More details
About this account
The agent searches retailers and deal guides against the established preferences and returns candidate offers.
Firsthand primary account opened and read; original publication date checked in primary metadata or visible timestamp. Outcomes are author-reported, not independently reproduced. Historical offers; no purchase or independent discount/feature verification. Original prompt excerpt reproduced exactly from the source; excerpt only, not the complete prompt.
Credit: Alex Hughes
Source published 12 Nov 2025
Source checked 15 Oct 2026
Tools and topics
ChatGPT agent browser · Retailer websites · Deal guides
Shopping · Reported deployment
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.