Codex to shop outfits from return history
Dini gave Codex a $500 budget to shop for a South of France vacation. Instead of asking for a generic list of fashionable clothes, she supplied a trip-specific Pinterest board and connected Gmail so the agent could learn from purchases and returns.
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Use my vacation mood board and purchase-versus-return history to suggest outfits within $500. Explain how each choice addresses the fit problems in items I returned.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Dini
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Dini reports buying the recommended vacation outfits within the $500 budget after roughly twenty minutes of agent work.
- Human involvement
- Dini made the purchases; the accessible preview does not describe a separate approval sequence for the recommendations.
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
- Provide the vacation budget and a Pinterest board for the intended style.
- Connect purchase and return history so Codex can identify fit patterns.
- Review the recommendations and purchase the selected outfits yourself.
Dini gave Codex a $500 budget to shop for a South of France vacation. Instead of asking for a generic list of fashionable clothes, she supplied a trip-specific Pinterest board and connected Gmail so the agent could learn from purchases and returns.
The board described the visual direction: pale yellow, white and light blue, with particular fabrics, necklines and silhouettes. Purchase history added a different kind of evidence. Clothes that looked appealing on a board were not necessarily clothes she kept. The agent could distinguish successful purchases from returned items and use those patterns when choosing recommendations.
The accessible account describes examples of those fit preferences. Very short dresses were rarely kept. Rigid fabrics and corsets without stretch caused problems, while bust and waist sizing could differ. Those constraints gave the agent something more concrete than an aesthetic label to work with.
Dini reports receiving recommendations within about twenty minutes and buying the recommended outfits within the budget. The public preview does not expose the full item list or independent purchase receipts, and its headline and body give different item counts. This case therefore does not repeat a count or invent selected brands. The suggested prompt reflects the verified inputs; the complete original prompting sequence is not available in the accessible preview.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: gurunandinisays.substack.com · Dini
Frequently asked questions
Why use returned purchases as context?
They revealed fit patterns that a mood board alone could miss, including rigid fabrics and very short dresses.
Did Codex complete checkout?
The author says she bought the recommendations. The preview does not establish agent-operated checkout.
Are the brands and item count verified?
The accessible preview does not expose the complete item list and gives inconsistent counts, so neither is reproduced here.
More details
About this account
Codex combined visual preferences with Gmail purchase and return history to recommend vacation outfits.
Firsthand accessible preview opened and read. Full prompting sequence and item details are behind the unavailable remainder; suggested prompt written by us.
Credit: Dini
Source published
Added to Agents In Use
Source checked
Tools and topics
Codex · Pinterest · Gmail
Vacation outfits · Purchase history · Fit preferences
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.