Feature Request: Context-Aware, Goal-Aware Reminders

Enable ChatGPT scheduled tasks/reminders to use recent conversational context to determine whether a reminder is actually needed, rather than triggering solely because a predetermined amount of time has elapsed.

More broadly, this could enable ChatGPT to function as an adaptive attention-management system: helping users keep important but non-urgent goals visible without requiring them to maintain a conventional task-management system.

The problem

Many important activities do not have natural deadlines.

Examples include:

  • professional learning and development;

  • maintaining relationships;

  • progressing long-term personal projects;

  • reflection and journaling;

  • strategic thinking;

  • health or lifestyle goals;

  • exploratory work and research.

These activities are easily displaced by urgent tasks, deadlines and whatever currently has the user’s attention.

Conventional task-management systems address this with recurring reminders: “Every two days, remind me about X.”

But this produces a second problem: the reminder fires whether or not it is useful.

For example, I want to maintain regular attention on an ongoing AI learning and reflection project. An appropriate instruction would be:

If I haven’t engaged with my AI learning for a couple of days, prompt me about it.

ChatGPT already has conversations with me that constitute that learning. I may have spent an hour discussing an AI use case, analysing an article, reflecting on an AI interaction, or developing my thinking.

In those circumstances, I don’t need a reminder.

However, the current scheduled-task capability cannot reliably evaluate that conversational activity before deciding whether to prompt me. The practical workaround is a recurring reminder every two days.

That loses much of the potential advantage of using an AI assistant rather than a calendar or task manager.

Why this matters

The problem is fundamentally one of attention management rather than task management.

People do not necessarily need more notifications telling them what is on their task list. They need help answering:

What deserves my attention now?

An AI assistant is unusually well positioned to help answer that question because it can potentially understand:

  • what the user is trying to achieve;

  • which goals are important to them;

  • what has a hard deadline;

  • what is important but has no deadline;

  • what the user has recently been working on;

  • what has quietly received no attention;

  • what has already been accomplished through conversation;

  • what is waiting on somebody else;

  • what the user currently has capacity for;

  • and what they are motivated to work on.

A static task-management system generally requires the user to explicitly maintain this information.

ChatGPT can potentially infer much of it through the user’s normal interactions.

Example use case

A user has an important professional-learning goal.

They know from experience that they will enthusiastically pursue it when it is in their attention, but may unintentionally neglect it when work deadlines, household responsibilities or other interesting projects intervene.

They ask ChatGPT:

“If I haven’t done anything related to my AI learning in a couple of days, remind me.”

Two days later, the scheduled task runs.

Before notifying the user, ChatGPT evaluates relevant recent activity.

Scenario A — no learning activity

There has been no relevant activity.

ChatGPT prompts:

“You haven’t touched your AI learning for a few days. Have you noticed anything interesting about how you or your colleagues have been using AI lately?”

Scenario B — substantial learning activity

Yesterday the user had a long conversation with ChatGPT analysing an AI implementation at work and recorded several new observations.

ChatGPT determines that the underlying goal has already received meaningful attention.

No reminder is sent.

Scenario C — partial activity

The user briefly mentioned an article they intended to read but did not return to it.

ChatGPT could prompt:

“You mentioned that AI workforce article a couple of days ago. Did you end up reading it?”

The reminder therefore becomes contextual and useful, rather than simply recurring.

Potential capability

Scheduled tasks could optionally be given access to an appropriate window of recent conversational activity and memory when evaluating whether their trigger condition has been met.

This could support instructions such as:

  • “If I haven’t progressed this project for a week, remind me.”

  • “If we haven’t talked about my learning goal recently, ask me about it.”

  • “Remind me about this unless I’ve already dealt with it.”

  • “If this project seems to have stalled, bring it back to my attention.”

  • “If I’ve been spending a lot of time on lower-priority projects while this important goal has received no attention, flag it.”

  • “Check in occasionally, but don’t bother me if I’m already making progress.”

The important distinction is that the scheduled event becomes an opportunity for ChatGPT to assess whether intervention is warranted, rather than an instruction to always generate a notification.

User benefit

This would allow users to preserve flexibility.

Instead of planning every week in advance or maintaining detailed project boards, users could tell ChatGPT:

  • what matters;

  • what must not be forgotten;

  • what deadlines exist;

  • and roughly how they prefer to work.

They could then continue working naturally.

ChatGPT could act as a lightweight safety net, selectively bringing important things back into attention when necessary.

This could be particularly valuable for users who:

  • manage many simultaneous projects;

  • work opportunistically rather than from detailed forward plans;

  • are highly motivated by current interest;

  • dislike maintaining task-management systems;

  • easily remember urgent work but neglect important non-urgent goals;

  • or want assistance balancing competing professional and personal priorities.

Why this is differentiated

Calendars already provide scheduled reminders.

Task managers already provide recurring tasks.

Kanban boards already track project status.

The distinctive opportunity for an AI assistant is not to reproduce those systems conversationally.

It is to combine context, memory, goals, recent behaviour and judgement to determine when the user would actually benefit from having something brought back into their attention.

In other words:

Don’t just remember the task. Understand when I need to remember it.

That would turn reminders from a scheduling feature into a genuinely AI-native attention-management capability.

Thanks for explaining the difference! Tasks can handle scheduled reminders and some monitoring: https://help.openai.com/en/articles/10291617. We'll share the request for reminders that recognize progress in recent chats and stay quiet when they aren't needed. No timeline to share.