OpenAI’s attempt to make ChatGPT a proactive, personalized daily companion has taken a sharp turn, with the company launching and then moving to shut down its new Pulse feature within weeks.

In September, OpenAI introduced ChatGPT Pulse as a mobile-only feature for paying Pro users. The tool was pitched as a way to “start your day” with AI-curated updates by letting ChatGPT learn from user transcripts and phone activity, including connected apps like calendar, email, and contacts.¹ The system would then perform personalized research overnight and deliver a morning digest via “topical visual cards you can scan quickly or open for more detail.”²

The launch fit into OpenAI’s broader strategy of building AI agents that move beyond reactive question-answering. One executive framed Pulse as part of a “big push toward AI agents,” an area where OpenAI and rivals like Anthropic and Google are investing heavily, envisioning assistants that can book travel, make reservations, and buy gifts on users’ behalf.² In this vision, Pulse was an early step toward assistants that understand user goals and act proactively rather than waiting for prompts.²

However, soon after the fanfare, OpenAI signaled a reversal. In an update tied to a revamped experience for scheduling tasks, the company said it was “sunsetting Pulse, which showed you custom daily digests in ChatGPT,” and that the feature would disappear “in the next 14 days.”³ Users were advised to rely on scheduled tasks instead for a daily briefing.³

From a product perspective, the move suggests OpenAI is consolidating around fewer, more flexible tools rather than maintaining a standalone digest product. For users drawn to Pulse’s promise of hands-off, personalized summaries, the shift may feel abrupt, but it also points to how quickly AI companies are iterating—and discarding—experiments in their race to define the role of everyday AI assistants.