Feature proposal
I use ChatGPT extensively in everyday life, including while traveling, shopping, choosing restaurants and wine, researching products, and making plans with family and friends. During a recent trip through Italy, several situations made me realize how much more useful ChatGPT could become if it were able to interact—with the user’s explicit permission—with other ChatGPT users and outside services.
I would suggest three related features:
1. Direct ChatGPT-to-ChatGPT sharing
Users should be able to authorize specific friends, family members, or contacts to communicate with their ChatGPT account.
For example, while traveling with my fiancée, I may ask ChatGPT about a restaurant, wine, attraction, or product and then want her to have the information too. Instead of creating a share link, copying text, taking a screenshot, or switching to Messages, I should be able to say:
“Alex, send this to Lorena.”
ChatGPT could then send the selected answer, photo, recommendation, or conversation segment directly to her ChatGPT.
Both users would have to approve the relationship, and users should have complete control over who can send them information.
2. Shared ChatGPT groups
Take that idea one step further and allow users to create shared ChatGPT groups.
For example, suppose 10–15 friends were traveling together. We could create an “Italy 2026” group. Everyone would voluntarily join, and ChatGPT would participate in the group.
I could then say:
“Alex, send the train information to the Italy 2026 group.”
“Send everyone the restaurant we found tonight.”
“Tell the group where we’re meeting tomorrow morning.”
The information would appear directly within the group’s ChatGPT experience.
More importantly, ChatGPT could maintain shared group context. Someone could later ask:
“What was the wine Ira recommended in Milan?”
“Which restaurant did they like in Amalfi?”
“What time is our train tomorrow?”
ChatGPT could answer using information that members had specifically shared with the group.
Private conversations and personal memory should remain private. Nothing should enter the group’s shared context unless the user explicitly shares it or has enabled clearly defined sharing permissions.
This could be useful far beyond travel: families, groups of friends, clubs, businesses, wedding planning, fantasy sports leagues, vacations, events, and many other situations.
3. Permission-based integration with personal applications and services
Users should also be able to voluntarily connect services such as Amazon, Spotify, retailers, travel services, loyalty programs, and other applications and specify what ChatGPT is allowed to access.
Amazon is a good example.
If I authorize ChatGPT to read my Amazon purchase history, I should be able to ask:
“Look at everything I purchased in 2026 and tell me where I could have saved money.”
ChatGPT could identify products I repeatedly purchase, calculate how much I’m spending, notice products I’m replacing unusually frequently, compare package sizes or subscriptions, and suggest alternatives that may last longer or provide better value.
Instead of simply recommending products based on a question I ask today, ChatGPT could make recommendations based on my actual purchasing behavior.
Spotify provides another example.
If I authorize access to my listening history, playlists, liked songs, frequently played artists, skips, etc., I could say:
“Create a three-hour playlist based on what I actually listen to, but make 25% of it new music you think I’ll enjoy.”
ChatGPT would understand my musical preferences from my actual behavior rather than requiring me to describe them.
The same concept could extend to many other services.
Privacy and user control
These features should be entirely permission-based.
Users should be able to control:
- Which people and groups can interact with them.
- Which information can be shared.
- Which applications ChatGPT can access.
- Whether an integration is read-only or can take actions.
- Whether ChatGPT must ask before sharing information or performing an action.
- The ability to revoke access at any time.
The objective isn’t to give ChatGPT unrestricted access to someone’s digital life. It is to allow the user to decide which parts of their digital life ChatGPT can use to help them.
Why I think this matters
ChatGPT is already extremely useful as an individual assistant. But people don’t live their lives inside isolated conversations.
We travel with other people. We shop. We listen to music. We make plans with friends and family. We belong to groups. We use dozens of applications that already contain information about our preferences and behavior.
Connecting those things—with strong privacy controls and explicit user permission—could move ChatGPT from being primarily an assistant that responds to questions into an assistant that genuinely understands and helps coordinate the user’s everyday digital life.
The interface could remain remarkably simple.
“Alex, send this to Lorena.”
“Alex, share this with Italy 2026.”
“Alex, look at my Amazon purchases this year and find ways I could save money.”
“Alex, make me a Spotify playlist based on what I’ve actually been listening to.”
Those are natural requests that users already understand without having to learn a new interface.
I think that would be an extremely useful evolution of ChatGPT.
That captures your idea without making it sound like a technical specification. More importantly, the Italy example and the Amazon example make the use case immediately understandable to somebody on the product team.