Last September, ChatGPT was a tool you opened when you had a question. Now it messages you first.
OpenAI's Pulse feature, which launched in September 2025, generates 5-10 personalized briefings overnight and delivers them when you wake up. It pulls from your conversation history, connected apps like Google Calendar and Gmail, and any feedback you've given it to assemble a morning summary. Think of it as a daily digest written by an AI that has read all your previous chats.
How Pulse Actually Works
While you sleep, Pulse runs asynchronous research based on your past interactions and connected services. It parses your email for important messages, checks your calendar for the day ahead, and surfaces information it thinks you'll want. The results show up as visual summaries designed to be the first thing you see in your ChatGPT app.
You can steer what Pulse focuses on by requesting specific automated reports or giving it feedback on what's useful and what isn't. The feature uses ChatGPT's Connectors system for app integrations and is available on web, with mobile rollout ongoing.
The Notifications Problem
Six months after launch, users are still being surprised when ChatGPT initiates contact. Social media posts from people startled by unprompted messages continue to pop up regularly. The issue isn't the feature itself, which is genuinely useful if you opt into it deliberately. The problem is the expectation gap: most people still think of ChatGPT as a tool that waits for input, not one that reaches out.
OpenAI had a related incident in 2024 where ChatGPT appeared to initiate conversations unprompted due to a bug, which the company quickly patched. Pulse is the intentional version of that behavior, but the lingering memory of the bug means some users can't tell whether they're seeing a feature or a glitch.
For anyone confused by a ChatGPT notification they didn't expect: check your Pulse settings. You can customize what it sends or turn it off entirely.