OpenAI is finally moving ChatGPT from a reactive chatbot to something that comes to you. Pulse delivers a personalised briefing every morning, learned from your chats, calendar, and Gmail. Here is what it does, what it does not do, and how to make it actually useful for real work.
ChatGPT Pulse is OpenAI’s first real attempt at proactive AI: each night ChatGPT scans your chat history, your saved memories, and (if you connect them) your Gmail and Calendar, then assembles a set of personalised research cards for the next morning. The feature shipped to Pro users in September 2025 and is rolling out to Plus subscribers across 2026. Most people set it up wrong and never use it again. Set it up the way I describe in this guide and it becomes the highest-leverage 5 minutes of your day.
Most professionals using ChatGPT use it the same way: you have a question, you open the app, you ask. That is reactive AI. You drive every interaction.
Pulse flips that. OpenAI describes Pulse as a feature where ChatGPT “proactively conducts research to deliver personalised updates based on your chats, feedback, and connected apps like your calendar.”[1] In plain English: overnight, ChatGPT looks at what you have been working on, what is coming up on your calendar, and what you have said is useful. Then it builds a set of visual cards waiting for you in the morning.
You wake up. You open ChatGPT. There is a Pulse tab. Inside are five to ten cards: a summary of what is coming up in your calendar, a research brief on the company you have a meeting with, a quick update on a topic you have been asking ChatGPT about, a draft of a reply to an email if you have asked for that pattern before.
It is not a news feed. It is a personal research assistant that has been working while you slept.
The feature shipped to ChatGPT Pro users (the $200/month tier) on 25 September 2025, currently mobile-only.[2] OpenAI has said the goal is to expand it to Plus subscribers and eventually free users as the system matures.[1]
Three things feed Pulse:
Your memory. If you have memory turned on in ChatGPT, Pulse draws from the facts ChatGPT has saved about you. Your role, your industry, the projects you have mentioned, the recurring themes in your chats.
Your recent conversations. What you asked yesterday, last week, last month. If you have spent two weeks asking about hiring frameworks, Pulse will probably surface a card about a new hiring trend.
Your connected apps. Right now this means Gmail and Google Calendar, with more on the way. If you have a meeting with Acme Co at 11am, Pulse can pre-research Acme and surface a one-page brief with the morning’s cards.
Each night, ChatGPT synthesises all of this and produces what OpenAI calls “focused updates” for the next day.[1] You give it thumbs up or thumbs down on each card. Over time it gets sharper. That feedback loop is what separates Pulse from a generic news app: it is not showing you what is trending, it is showing you what is trending for someone in your role doing your work.
The mental model: Pulse is what a really good chief of staff would put on your desk every morning. Not the news. Not your inbox. A curated set of things that map to what you are actually working on.
Here is the honest truth. I have watched a lot of people try Pulse and then quietly stop using it within two weeks. The pattern is almost always the same.
They turn it on. They do not configure memory. They do not connect Gmail or Calendar. They look at the cards for three mornings, get vague AI-news headlines and generic productivity tips, and decide Pulse is fluff.
That is not Pulse failing. That is Pulse not having any signal to work with.
The feature has almost no value if ChatGPT does not know what you actually do. If your chat history is “write me an email to my landlord” and “recipe for banana bread,” Pulse will give you nothing useful.
Pulse is only as good as the data you feed it. The 5 minute setup most people skip is the difference between “this is a gimmick” and “this is my favourite ChatGPT feature.”
Do this once and you will get value every morning.
Step 1: Switch memory on and give it real raw material. Go into ChatGPT settings and confirm memory is enabled. Then open a chat and tell ChatGPT exactly what you do. Use a prompt like this:
“I want you to remember the following about my work. My role is [your title]. I work in [industry]. The projects I care about right now are [3 to 5 specific projects]. The metrics I am measured on are [2 to 3 metrics]. The tools I use daily are [3 to 5 tools]. The topics I want to stay current on are [3 to 5 topics]. The kind of meetings I have most weeks are [types]. Save all of this to memory.”
This is the highest-leverage prompt you will ever write. It is the seed Pulse grows from.
Step 2: Connect Gmail and Calendar. In ChatGPT settings, find Connectors and authorise Gmail and Google Calendar. Yes, this is a privacy decision. Read the data-use policy. If your role involves regulated client data, talk to your IT team first.
Step 3: Train it for 5 mornings. Open Pulse. Thumbs up the cards that are genuinely useful. Thumbs down the ones that miss. Tap the “curate” option and tell it directly what you want more of and less of. The first 5 days are noisy. By day 7 you will start seeing genuinely good cards.
Step 4: Give it a weekly direction. Once a week, in a normal chat, tell ChatGPT what you want more of next week. For example: “This week I want Pulse to focus on AI marketing news, my Tuesday board meeting, and any updates on the [Project X] launch.” Pulse will reweight accordingly.
That is it. The whole setup takes one cup of coffee. Most people skip it and then complain Pulse is shallow. Do not skip it.
These are the patterns I have seen pay off across a few hundred professionals testing Pulse in our workshops.
1. Pre-meeting briefings. If your day has 3+ meetings and you connect Calendar, Pulse will quietly pre-research the people and companies on your day’s schedule and surface a one-page brief. This is the single most valuable use case for client-facing roles (sales, consulting, account management). What used to be a 15-minute LinkedIn-and-news scan before a meeting now happens automatically.
2. Industry surveillance. If you tell Pulse you care about, say, the future of work in financial services, it will quietly pull together what changed yesterday: a regulatory update, a competitor announcement, a new study. Better than a Google Alert, because it filters for what is actually relevant to someone in your seat.
3. Project status threading. If you have been working on a specific project across multiple conversations, Pulse can produce a daily “where you left off” card. It is not perfect. But for people running multiple parallel projects, it is a memory upgrade that you previously had to assemble manually.
4. Inbox triage signals. Pulse cannot read your email back to you, but it can flag if something time-sensitive arrived overnight that maps to a project it knows you care about. Not a replacement for inbox triage. A pre-filter.
For a deeper look at how to build daily systems around AI tools like this, our personal AI workflow system guide walks through the whole stack.
Let me be straight about the things Pulse does not do, despite the marketing.
It does not replace email triage. Pulse can flag things, but it cannot reply to your inbox for you, and the surfaced cards are not granular enough to clear an inbox.
It does not replace your actual prep work. The pre-meeting brief is good. It is not the document you walk into a strategic client meeting with. Treat Pulse output as a starting point, not the finished work.
It is not free. As of May 2026, Pulse is available to Pro and Plus subscribers. If you are on the free tier, you do not have it yet.[2][3]
It can be wrong. Like any LLM-based system, Pulse can produce a confident card based on stale or incorrect information. Verify before you act on anything specific. For more on why AI gets confident-sounding things wrong, our anti-hallucination toolkit covers the patterns.
It is mobile-first. The cleanest experience is on the ChatGPT mobile app. Desktop support is improving but lagging. If you are someone who lives in browser tabs, this might feel awkward at first.
The privacy trade-off is real. Connecting Gmail and Calendar to an AI system that pre-processes them every night is a meaningful decision. Make it on purpose, not by reflex.
Three concrete moves.
If you have Pro or Plus: Run the 10-minute setup above today. Do not skip the seed prompt. Connect Calendar at minimum. Give it the rest of this week to learn, then judge it on day 8, not day 1.
If you are on the free tier: Pulse is not available yet, but the underlying behaviour (using memory and feeding ChatGPT a strong context prompt about your work) still pays off in every other interaction. Go set up memory now. When Pulse rolls down to your tier later in 2026, your account will be ready.
If you manage a team: Test Pulse on yourself for two weeks before recommending it. The hardest part of this kind of feature is that the value depends entirely on configuration. A bad rollout where everyone gets thin cards will turn your team off proactive AI for a year. A good rollout where each person is shown how to seed memory will land. Pick three early adopters, give them the setup guide, gather feedback in two weeks before going wider.
Proactive AI is the direction every assistant is moving in. ChatGPT Pulse is the first mainstream example. Anthropic’s Orbit and Google’s Gemini proactive features are right behind it.[4] Learning how to configure one of these well now is a skill that compounds across every tool you adopt next.
ChatGPT Pulse is a proactive feature from OpenAI that delivers a personalised set of research cards each morning. It draws on your chat history, saved memory, and connected apps like Gmail and Google Calendar to produce updates relevant to your work. It was launched to Pro subscribers in September 2025 and is rolling out to Plus subscribers across 2026.
Pulse launched on the ChatGPT Pro tier, which costs $200 per month. OpenAI has confirmed it is expanding the feature to Plus subscribers ($20 per month) and plans to make it available to free users over time. As of May 2026, Pulse is not available on the free tier.
Only if you choose to connect Gmail and Google Calendar in your settings. Without those connectors, Pulse only uses your chat history and saved memory. Connecting Gmail allows Pulse to use email context to inform briefings. The data-use policy is documented in ChatGPT settings and should be reviewed before connecting any work email.
It depends on whether you take 10 minutes to set it up properly. Pulse without memory and without connected apps is shallow and most users abandon it. Pulse with a seed prompt about your role, memory enabled, and Calendar connected becomes a useful pre-meeting and project surveillance tool, especially for client-facing roles.
No. Pulse is a pre-filter and a research assistant, not a replacement for email triage or planning. It works best as a 5 to 10 minute scan that surfaces what to dig into next, not as the place you actually do your morning work. Treat its output as a starting point.
This guide is part of Future Factors AI’s ongoing effort to make AI useful for non-technical professionals. Written by Sana Mian, Co-Founder of Future Factors AI, an AI training company that has helped 2,000+ learners build practical AI skills through bootcamps, corporate workshops, and keynote sessions. Visit our AI Courses page to learn more.
Weekly tips. Real examples. Practical help for busy professionals.
We care about your data, check out our privacy policy.
