Inside OpenAI’s New Scheduled Tasks Page and the Push for a Proactive Superapp

Instead of a separate feature, OpenAI is directing users to ask ChatGPT to schedule a daily briefing based on their interests and chat history. The capability Pulse offered—proactive, personalized research—now lives inside the more flexible scheduled tasks system, where users control the timing and content of their briefings .

How natural-language tasks work now

Users create tasks the same way they talk to ChatGPT: in plain English (or any supported language). You can ask it to send a reminder at a specific time, compile a weekly report, monitor a website for changes, or deliver a morning news roundup every weekday .

Tasks support both one-time and recurring schedules. You can pin a task to a precise time or a broader window—morning, afternoon, or evening—and monitoring tasks can search the web and check for changes on your behalf . While the full set of capabilities will grow, the core workflow is now accessible without switching to a separate model or buried menu.

The long road from January 2025 beta

The Tasks feature first appeared in beta on January 14, 2025, restricted to Plus, Pro, and Team subscribers . It was a limited experiment: users had to manually select a “GPT-4o with scheduled tasks” model and could only set up to 10 active tasks . There was no central management page, and reliability was inconsistent.

OpenAI’s push to make ChatGPT more proactive accelerated throughout 2025 and 2026. In September 2025, Pulse appeared as a Pro-only preview that researched overnight and delivered morning briefing cards . By early 2026, Fidji Simo, OpenAI’s CEO of Applications, publicly declared that the year’s goal was turning ChatGPT into a “true personal super-assistant” that remembers context and acts without prompting .

The underlying models also matured. OpenAI released its o3 and o4-mini reasoning models in April 2025, optimized for fast, cost-efficient reasoning in coding, math, and visual tasks . These models form the engine behind the more reliable scheduled task execution users now have.

Why this matters: the superapp strategy

Scheduled tasks are not just a productivity feature; they are the proactive foundation for OpenAI’s planned superapp overhaul. Throughout the first half of 2026, multiple reports revealed that the company is preparing to merge ChatGPT with its Codex coding platform and Atlas browser into a single desktop application . The goal is to funnel 900 million users toward paying products—especially Codex—ahead of an anticipated IPO .

In a June 2026 statement, Thibault Sottiaux, who leads OpenAI’s core product and platform, described working toward a product where users stay inside one experience for everything . OpenAI’s own enterprise strategy page frames it as a “unified AI superapp” where frontier AI governs all agents and employees get work done .

Scheduled tasks play a specific role here: they make ChatGPT an active, background participant in a user’s day. Instead of waiting for a prompt, it pushes reminders, monitors conditions, and delivers scheduled reports. Pulse’s sunset confirms that OpenAI sees a single, consolidated task engine as the right architecture for this proactive layer.

With the June 2026 update, the feature transitions from an optional beta into a central pillar of how ChatGPT will operate as a superapp. It is no longer just a chat interface that can schedule things—it is becoming a scheduled agent that can also chat.