Feature Request: A More Personal, Proactive, and Flexible ChatGPT
I use ChatGPT frequently, and as I use it more deeply for learning, projects, planning, and everyday tasks, I increasingly feel that its next major step could be moving beyond a chatbot that waits for questions.
I would like to share four related product suggestions. These are mainly about the direction I would personally like ChatGPT to evolve toward rather than detailed implementation proposals.
1. Turn the Desktop Experience into a Personal AI Workspace
I would like ChatGPT’s desktop and agent experience to gradually become a unified entry point for things such as:
- Calendar and schedules
- Tasks
- Projects
- Knowledge and notes
- Files
- Connected third-party services
Today, a user’s information is often scattered across calendars, Notion, task managers, local files, note-taking software, and ChatGPT conversations.
This means that every time I want AI assistance, I often have to find the relevant information, provide it again, explain the context, and tell the AI where the project currently stands.
I would prefer a model where ChatGPT can understand these different parts of my digital life as one connected workspace.
Third-party applications can continue to store and manage the actual data, while ChatGPT becomes the intelligent layer that helps me understand, organize, plan, and act on it.
For example, instead of only asking:
«“Create an event for tomorrow.”»
I should eventually be able to say:
«“I need to finish this project this week. Look at my current progress, tasks, and calendar, and help me rearrange the remaining work.”»
That feels much closer to what a personal AI agent could become.
2. Expand Study Mode into a Long-Term AI Tutor
I would also like Study Mode to evolve into a much more complete personal learning system.
This could work with the user’s own learning materials in a way similar to NotebookLM-style source-based learning: textbooks, PDFs, lecture slides, notes, exercises, and other course materials could become the knowledge base for a subject.
ChatGPT could then help users learn very different types of subjects, such as:
- Languages
- Mathematics
- Physics
- Accounting
- Programming
- Design
- Blender, Unreal Engine, and other professional skills
The important part is not only explaining concepts.
I would like the AI to be able to:
- Understand what I am currently learning
- Remember my learning progress
- Generate exercises and tests
- Identify weak areas
- Review mistakes
- Help me revise previously learned material
- Adjust the learning process based on my performance
Different subjects also require different teaching methods. Learning mathematics should not work exactly like learning English or learning Blender.
The goal would be for ChatGPT to gradually understand not only what I want to learn, but also what I already know, what I struggle with, and what I should learn next.
In other words, I would like ChatGPT to become a persistent AI tutor rather than only an AI that answers individual study questions.
3. Give ChatGPT Proactive Intelligence
This is different from scheduled tasks or ordinary reminders.
Today, most AI interaction follows this pattern:
The user notices something → the user asks the AI → the AI responds.
I would like ChatGPT to eventually also support:
The AI understands the context → notices a relevant risk, need, or follow-up → proactively helps the user.
For example:
After I finish learning something, ChatGPT could later remind me that it may be time to review it.
After a learning session, it could follow up and ask whether the material was difficult or whether I need more practice.
If severe weather such as a typhoon may affect me, it could proactively tell me what preparations may be useful.
If my calendar or project information suggests that a deadline may become difficult to meet, ChatGPT could warn me before the problem actually happens.
If a scheduled activity may conflict with weather, another event, or project progress, the AI could point that out.
The important difference is that I may not already know that I need to ask.
This would move ChatGPT from simply responding to requests toward helping users anticipate problems and needs.
At the same time, users should remain in control of this behavior.
I should be able to decide which types of things ChatGPT is allowed to actively notify me about, which things should appear silently, and which things should not generate notifications at all.
The exact implementation is something I would leave to the product team. The important idea is that proactive AI should exist, while the user should still control how much the AI is allowed to interrupt them.
4. Offer More Flexible Subscription and Usage Options
My final suggestion is about pricing and usage limits.
There can be a large gap between a normal subscription tier and a much higher-usage tier.
My personal problem is simple:
1× usage may not be enough, but 5× usage may be much more than I need.
I would be willing to pay more for something like 2× or 3× usage, but I may not necessarily need all of the additional premium features included in a more expensive plan.
For example, some users may simply want:
- More usage of the models they already use
- Stable models rather than access to every newest frontier model
- Higher limits without additional premium tools
- The ability to pay for more usage without upgrading to a much larger bundle
Personally, I may not need features such as deep research or every advanced model. Sometimes I simply want more usage of the stable models I already rely on.
I think it would be useful if OpenAI could consider more flexible options, such as intermediate usage tiers, additional usage capacity, or more configurable combinations of usage and features.
The exact pricing model is naturally something for OpenAI to decide.
The main point is:
Needing more usage should not necessarily mean needing to buy every additional feature.
The Overall Direction
All four suggestions point toward the same broader idea.
I would like ChatGPT to gradually evolve from an AI that primarily waits for me to ask questions into a personal AI system that can:
understand my context, help me learn, organize my work, anticipate relevant needs or risks, and assist me at the right time.
At the same time, I want to remain in control of my data, notifications, features, and how much I pay for the capabilities I actually need.
I think that combination — stronger long-term context, better learning support, proactive assistance, and more flexible pricing — could make ChatGPT much more useful as an everyday personal AI.
Thank you for reading.