Hi Team OAI and everyone 
I use ChatGPT heavily as a normal consumer, not as a developer or enterprise customer, and I think one of the biggest product problems right now is that very different user groups are still being served inside essentially the same product philosophy.
I would strongly separate ChatGPT into different experiences:
1. Youth space Children should not simply receive a watered-down version of the adult product. An 8-year-old and a 16-year-old also need different levels of access.
A youth account could include parental account linking, age-appropriate permissions and learning-focused safeguards. ChatGPT should help students understand homework, explain concepts, review their own work and show examples, but it should not simply complete school assignments for them to submit as their own.
Parental controls should be granular rather than all-or-nothing. Sensitive features could require parental permission. Serious safety concerns could have an appropriate escalation design without turning every private conversation into automatic surveillance.
2. Adult consumer space Verified adults need a genuinely adult environment.
This does not mean “anything goes.” It means safeguards that understand context and actual risk rather than treating adult conversations as if every user were a child or in crisis.
Adults should be able to discuss relationships, sexuality, mortality, difficult emotions, dark humor, health concerns, art, politics and other complex topics without the assistant constantly becoming sterile, alarmist or overly paternalistic.
Adult ChatGPT should also be more creative, expressive and flexible. A normal consumer AI companion should be good at writing, brainstorming, everyday problem-solving, fitness, interpreting bank letters or contracts, budgeting, workplace questions, health information and daily planning — not immediately browse the web or jump to the worst medical possibility whenever a user mentions a symptom.
3. Coding / Enterprise space Coding and enterprise users obviously need additional compute, tools, security controls and higher reasoning tiers. But consumer users should still have access to every major new model, as they do today, even if enterprise accounts receive higher usage limits or additional reasoning modes.
Being a non-developer does not mean a person only has simple problems.
Then I would add a number of consumer-focused features that are currently missing:
Explicit image-generation control: a setting such as “Generate images only when I explicitly ask.” No automatic triggering from descriptive language.
Delete generated images independently: users should not have to delete an entire conversation just to remove an image from their image library.
Better consumer connectors: OneDrive, Reddit, X, Facebook and Instagram would be highly useful for ordinary users.
Source-controlled scheduled news: allow users to schedule a daily briefing while specifying not only topics but exact preferred sources or accounts.
Alarm-triggered routines: when I dismiss my phone alarm, ChatGPT could run a routine in a conversation I selected and send the first message there — for example weather, calendar, important email, reminders and fitness plans. It should not be forced into a separate automation chat.
File upload choice: whenever a file is attached, ask whether it should remain only in that conversation or also be permanently stored in the Library. Automatically duplicating everything into the Library creates clutter.
Persistent model selector: always display the active model and make switching models obvious instead of hiding the selector.
Full conversation export: export the entire conversation, including both user and assistant messages, to PDF, DOCX or another readable format.
Search inside conversations: search both user and assistant messages in long chats.
Pin conversation sections/messages: allow important moments to be bookmarked and filtered later.
Better mobile memory transparency: on mobile, show which memories, custom instructions, files or other sources influenced a response, and allow the user to edit/delete/rate those memories directly there.
Per-chat memory controls: decide what a specific conversation may remember.
Per-chat browsing preference: e.g. “only when asked / when necessary / automatically.”
Better Hungarian Voice Mode: current Hungarian speech can sound noticeably non-native and unnatural.
Real visual customization: themes, backgrounds, fonts, chat colors and more expressive interfaces instead of a single corporate visual style.
Customizable sidebar: users should be able to prioritize the tools they actually use and hide product areas that are irrelevant to them.
Visible update communication: ordinary users should not have to follow Reddit, X or support pages to discover major changes. Add an in-app “What’s new” popup and/or optional update email explaining new features, removed features, model changes and important behavior changes in plain language.
The broader point is simple:
Consumer AI should be designed around how ordinary people actually live with AI every day, not as a simplified version of a coding or enterprise product.
A large portion of ChatGPT’s value happens in long-term everyday use: conversations, creativity, learning, relationships, organization, health and fitness, personal projects and daily routines. Those users deserve product design built specifically for them too.
Thanks