Agents In Use
What’s new todayLast update
47 use cases added today

ChatGPT plans a group weekend and builds a coordination site

Sam Altman asked ChatGPT Work to plan three weekend options for nine friends, build a coordination site, and prepare an email after the group chose a destination.

Source published x.comAgent: ChatGPT agentvia ChatGPT Work
FROM THE SOURCE

Original prompt

Shared by the source author. See source notes for context.

use all my chat history to figure out ideas for a long weekend trip with 8 friends

Evidence at a glance

Sourced account
Source poster / publisher
Sam Altman (@sama)
Original account
Read the source ↗
Published
Source link checked
Reported result
Altman reported that the planning, site-building and email-drafting workflow worked. The source does not confirm a group decision, live public site or completed reservation.
Human involvement
The group was meant to choose a destination before reservations. The post does not say whether the site was shared, whether the group agreed or whether any booking was completed.

A linked account supports attribution. It does not mean we independently reproduced its result.

Report an issue with this caseOpen correction email

Opens your email app with the case link included. Nothing is sent automatically.

How the agent did it

  1. Altman asked ChatGPT Work to use his chat history and produce three long-weekend options for a group of nine.
  2. He asked it to build a site where the friends could compare activities and decide together.
  3. He specified that reservations should wait until the group agreed, and asked for a Gmail invitation draft once the site was ready.

Planning a weekend for nine people means more than picking a destination. Sam Altman asked ChatGPT Work to use his chat history to suggest three options, build a small site where the group could compare what to do in each place, and wait for agreement before making reservations. He also asked it to draft a Gmail message for inviting the group once the site was ready.

The important sequence is the approval gate: research and comparison first, group decision second, reservations afterward. That keeps the agent from booking a trip before the people involved have agreed. Altman said the workflow “just worked,” but his post does not name the destinations, show the site, confirm that friends used it, or say that any reservations were actually made. A reply joked that the site was only available at a local address; that is a commenter’s inference, not something Altman confirms, so the post alone cannot establish that the site was publicly accessible.

The shared prompt excerpt below is from Altman’s original request. To adapt the idea, ask for options with a clear comparison, make the group’s approval an explicit condition, and check every price and reservation before committing.

Prompt
Original prompt

Source: X · Sam Altman (@sama)

Frequently asked questions

What did ChatGPT Work make?

Altman asked it to compare three long-weekend options, build a site for nine friends to coordinate, and draft a Gmail invitation.

Did the group choose a destination or make reservations?

The post says reservations should happen after group agreement, but does not report that the group agreed or booked.

Was the coordination site publicly accessible?

The post does not say. A reply speculated it used a local address, but that is not confirmed by Altman.

More details

About this account

ChatGPT Work researched options, created a coordination site and drafted an email; the requested reservation step was conditional on group agreement.

Reviewed archived direct X capture (HTTP 200, 2026-10-15); current web reader returned 403. Original prompt excerpt: “use all my chat history to figure out ideas for a long weekend trip with 8 friends”. It does not establish that the site was public or reservations completed.

Credit: Sam Altman (@sama) — X

Source published

Added to Agents In Use

Source checked

Tools and topics

ChatGPT Work · Gmail

group travel · trip planning · coordination site

Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.