Connecting ChatGPT Pulse to my Gmail to Test It was a Terrible Mistake. Here's what Happened.

The privacy concerns are real, and the results were mixed; I'm not sure what I just agreed to.

---

Source: OpenAI

OpenAI released an update called ChatGPT Pulse a few months ago. It’s literally ChatGPT that will accompany you every morning with personalized updates based on your chat history, calendar, and connected apps. But is this genuinely an effective productivity enhancement, or is it just another way to make you dependent on an AI subscription you’ll eventually forget why you’re paying for?

I connected my entire Gmail and Google Calendar to test it live.

Let me tell you what happened, what worked, what failed spectacularly, and why this feature reveals more about AI’s future than you might think.

Why AI Companies Are Competing on Ergonomics, Not Power

Let’s be honest: AI performance is stagnating. Whether it’s a lack of computing resources, financial problems, or market saturation, the raw capabilities of generative AI and large language models are hitting diminishing returns. So, we’re seeing each company test minor features to differentiate themselves.

ChatGPT Pulse is now available in preview and in beta in Europe since the end of October. It’s a fresh experience where ChatGPT can proactively provide daily updates personalized from your chats, your feedback, and your connected applications, like your calendar. Rolling out today to Pro users on mobile.

The promise, let me explain it in simple terms, is that based on what you tell ChatGPT the night before, based on what ChatGPT knows about you from your history and its memory feature, and from what ChatGPT has connected, like your calendar and other sources, it can suggest personalized content every morning.

Why morning? Every day, a notification center delivers three items, including articles, content, and information, designed to be relevant to your upcoming experiences.

My Honest Take: Cool Concept, Questionable Necessity

On one hand, I find this cool, but I don’t really have a use for it. But I think I could definitely use it in my work and personal life. For example, if it sees I have a dentist appointment and tells me, “Here are three articles to prepare for your dentist visit” or “This scientist made an incredible discovery about cavities,” that’s going to annoy me.

Or if the night before I told it, “Yeah, I have problems with my accountant, I need to do this and that because of my business taxes,” and the next day it pushes articles about business taxes, that could be cool. But honestly, I don’t really know.

It needs testing.

However, for work, I can see myself asking it every Monday morning, every Wednesday, and every Friday to send me the five tech, design, and graphic news items that dropped. I’d be super happy in the morning, drinking my coffee, checking things quickly.

What I wouldn’t like is push notifications. I hate those. But opening ChatGPT in the morning and having it say, “Hey Novy, I know you want your tech news, and economy, here they are. By the way, I also saw you have a trip to Shanghai, so there’s that. I also saw you had this.” I’d be okay with that for work purposes. I’d be okay, having an assistant like that.

I’m telling you right now, I could be a big customer for this, with all the privacy concerns that entails about giving information about yourself. Your account needs to be well secured, and so on.

How do you activate it? Today, it’s in preview for Pro accounts. I checked this morning, wanting to have it, and I didn’t. Unfortunately, I’ll retest. It was only mobile for the preview. However, it’s also on the web now.

If you have it, there’s a tab at the top of your side menu. Pro users — that’s the $200 plan, not the $20 one.

So it’s for those paying $200, which means I won’t have it.

Essentially, it’s a comfort feature with pros, but in my humble opinion, it will roll out to everyone. Why? Because honestly, it doesn’t require many resources. You could already have this. Every morning, you could copy and paste a prompt saying, “based on today’s news and what you know about my calendar and conversation history with you, generate five recommendations,” make a nice prompt that you copy each time, and manually request your thing.

The idea is having ergonomics where you don’t have to do it.

The Addiction Trap Nobody’s Talking About

That would interest me for professional use, but my take is that this is yet another thing to get your foot in ChatGPT, a foot in a service, a foot in a social network, a grip, an addiction, an even robust ingrained habit so you can no longer do without it and renew your subscription again and again and again without really knowing why you’re subscribed anymore.

But overall, we can’t throw stones at OpenAI or anyone else. It’s pretty much all web businesses and social networks.

Don’t 12,000 ChatGPT wrappers already do this? Yes. Obviously, Perplexity, this plugin, this wrapper, this free thing already does it. The answer is always yes. But a little video with little sounds, cute little ergonomics with little cards, little notifications connected to your history, and ChatGPT memory — all that, you’re paying for the comfort of an ecosystem.

It’s always that. Yes, but I don’t buy that because it’s free elsewhere, but they did it well. They charge for it, and I pay for things like an iPhone, for example, and so on. You need to be part of the elite to afford $200 per month.

I think we’ll see later, but I think it will be in place everywhere because it’s not that expensive for them. Whether they want to keep a tiered effect for features like this, I don’t know. But right now they’re putting it for pros because they’re in beta.

I Connected My Gmail and Calendar. Here’s what actually happened.

Free City already has a daily news module for everyone. Absolutely. But that’s not the problem with daily news. That’s not the promise. It’s personalized news. But personalized to a point where it knows your calendar, it knows the discussions you had with AI the night before.

In personalization, with more information for you, you can ask ChatGPT to give the news every day. It’s not super optimized, but it works okay. Yeah, that already exists. But this is a packaged version, let’s say, cute little cards, little notifications, and everything.

The same as you.

It’s cool for professional use, but there’s a mega warning about addiction. I also wonder if creativity and reactivity won’t suffer. For professional use because I do my monitoring manually every day, and I sometimes research on X, on Grok, and blogs. I go searching, keep searching until I find the best news.

That’s acceptable if it just saves me 10 minutes instead of starting a manual search. AI, for now, has power problems.

Because I’d rather choose maybe later my AI assistant and LLM, whatever, pretty because it can log into my calendar and I tell it “yeah, add an appointment such-and-such and check my availability and send to X such-and-such” rather than something super performant.

No, I’d rather be on everything that’s API, all the interfacing between objects, between applications. I’m waiting for the moment when it scolds you because you haven’t done anything in 3 days.

I hope so.

The Live Test: Connecting Everything to ChatGPT

So MCPs, thank you. This allows interfacing LLMs with software. For example, yeah, tell Dylan I’m 15 minutes late and add a slide to my Google Slides to thank the client. Can you imagine that? You don’t need much computing power, but you need interfacing, APIs, MCPs, and that’s what’s going to be more interesting in the future for me.

Why not? I always have lots of notes the night before for the next day, and ultimately, lots of things fall through the cracks because of forgetfulness for now. Laziness, if the thing optimizes me, maybe it’s interesting.

I’d love it, you know, sometimes you take notes in the evening when you have lots of ideas before going to bed after your shower or when you walk your dog. You speak into the chat, and you say, yeah tomorrow remind me and format that nicely.

The research process itself is a creative exercise. It’s notable that the pre-chewed part I’m talking about.

The question is resource renewal. Yeah, but anyway, it doesn’t bother me if it sends me lots of news and lots of articles, and the ones that interest me, I dig deeper. It remains research. Anyway, what’s cool with ChatGPT and AI is that it knows. I can tell you know my name is Novy from Nov Tech, and I’ll find articles that will be interesting.

Pulse, a personalized morning briefing, is launched by ChatGPT. Pulse offers each morning a set of unique cards with news, reminders, and practical ideas to transform ChatGPT into a morning routine. And that’s what’s interesting, for it to be your little morning cigarette.

Currently reserved for Pro plan subscribers at €230, but it’s not limited to professional tasks. For some users, ChatGPT has now become a reflex much more than an occasional tool or a fantasy. OpenAI’s chatbot has become for many a daily assistant capable of explaining administrative papers, summarizing current events, or simply answering a general culture question, replacing Google, of course.

And this reflex doesn’t displease OpenAI, which wants to make its flagship product essential. It’s with this in mind that the firm launches Pulse. Its objective is that rush to ChatGPT before their first coffee.

To create automatism, a habit. Pulse presents itself as a daily appointment with ChatGPT. Each night, the tool analyzes the user’s preferences, their conversation history, and if they wish, applications they’ve connected to, like Gmail or Google Calendar.

Source: OpenAI

I already said it a bit, but it’s a real shortcoming for me at the moment with ChatGPT in the Plus version that it doesn’t know my emails, my calendar, and everything. So it’s true that it can be scary for it to know everything, but it can also be a genuine comfort.

It could be the point that makes me go toward Gemini one day, but well, I don’t really know how to position myself on that yet. Upon waking, it then offers between 5 and 10 synthetic sheets called cards that group essential information deemed relevant.

These cards can take several forms: a summary of articles on a regularly followed subject, dinner meal suggestions, a travel itinerary idea, or reminders linked to the schedule. The user can delve deeper into a topic by opening the corresponding card, asking for supplements, or simply moving on to something else.

Each update is only available for the day unless the user chooses to save it. Pulse also stands out for its deliberately limited approach. Unlike news feeds or social networks, it doesn’t seek to retain attention indefinitely. Once the day’s information is delivered, the application displays a closing message: “That’s all for today.”

The Privacy Concerns I Discovered During Testing

Well, I know, but in itself, for professional use, it’s cool. It’s that for professional use, you want to see your news, and you don’t want to be told, you know, sometimes I look for news on Instagram, Twitter, and then I find myself scrolling too long, and I know I won’t find anything more.

Even if it’s creating a habit, an addiction, at least there’s that. Honestly, I’d really like to test to see how accurate it is or not. The system can be refined by the user. It’s possible to give positive or negative feedback on received cards or even explicitly request certain content for the next day.

Because of this progressive personalization, suggestions become more relevant over time. I really want to test, don’t you? I plan to test how it could work. What could it bring me? I admit it works on me, marketing-wise.

For now, it’s exclusive for their pros, available only on mobile, and to subscribers of the Pro plan at €230. A choice that doesn’t seem guided by the features of nature. It’s not expensive. If it can accompany professional objectives, it’s equally suited to the most basic daily tasks.

This choice could be motivated by the large amount of resources used to generate these reports every day. Really? For me, it’s rather low on resources. It looks at history and searches for articles. I mean, in pro mode, you can do advanced searches. So, that doesn’t seem that expensive to me. Maybe doing it every day is costly.

With ChatGPT, you open it sometimes not for several days, and yeah, maybe. I don’t know. I would have liked to have a pro’s life, but for me it’s not very expensive. But probably I’m wrong.

The Moment I Actually Connected My Gmail

It would need to be really well-chosen news, but I believe in it. Am I going to crack and buy a €230 subscription? No, no, I’m not into that. I don’t want to. But I think it can be good. Yeah. It also avoids consulting articles on sites polluted with advertising.

So you pay your subscription, but you don’t have ads in articles. But these articles feed the AI.

How are we going to figure this out? At some point, ChatGPT will have to pay for the articles it absorbs because if there are ads on these articles, it’s because you have to pay the editorialist, you have to pay the journalist.

I imagine it’s expensive if they have to do it for all subscribers who activated Pulse over time. Actually, I think it’s costly because if it generates for nothing and people don’t consult it, yeah, but if you let it be activated manually, maybe it would be okay.

OpenAI wants to improve Pulse’s effectiveness before widening access — first to Plus subscribers, for example, me, then to all users. Long-term, OpenAI displays its ambition to design an AI system capable of researching, planning, and taking actions for you according to your directives so you can progress.

I agree with this vision of AI. I told you earlier, but this can somewhat cross-use my tools securely. Yes, since 2025, OpenAI has introduced what we call connectors, a Plus subscription. Okay, I can do it. Connect Gmail.

But I did it.

Once connected, I’ve been able to access my emails and my events with my permission to answer your queries like, “What do I have today?” My dream — I’ll do it. There’s also a deep search function that allows searching within your Gmail.

Oh yes, settings, connector.

That’s it, we can.

Before, it was only for pros. Google Drive, too. I just canceled my Google Drive for the C drive. Do you think that’s serious?

What am I doing? Is it serious? I’m giving OpenAI access to my Gmail and my Google Calendar. What do you think? I was saying to myself: Am I doing something stupid, or is it okay?

Yes, it’s stupid.

It made me tremble a bit, but I want to, you know. Ouch.

It was like vaping.

Nobody knows if it’s serious. If it’s only professional, it’s fine. Yes, it’s only professional. But well, sometimes I put an appointment with the proctologist on my calendar.

It asked me which calendar I should activate. Alright, go ahead. I do this.

ChatGPT accesses your Google account. What could go wrong? Google, OpenAI, take everything. What could go wrong? Google Calendar is now connected.

What could scare me is my life.

Gmail and Google Calendar are now connected. Now, in chats, you can reference your emails and your calendar events in your responses.

Try, “What email should I respond to?”

Wait, I’ll try. I swear I’m trying, but I can’t show you because it’s my mailbox, sorry.

Can you show me the emails you’re hesitating to process?

ChatGPT is stupid. It’s ridiculous as hell. It’s as stupid as a post. Bro, you have access to my Gmail now. No, no, it told me no.

What a fool.

Oh, I need to restart the tab.

I had to refresh. Discussion with Gmail. It is written stylishly and honestly. I think it’s much more stylish in voice mode. Oh yeah, I can’t show.

Source: OpenAI

Wait, there was a problem.

It was showing me emails that were read 10 years ago. Oh, I didn’t specify unread emails because I think I didn’t respond.

Oh no, it was a fiasco.

I’ll explain what happened. I clicked on open email, and it opened another email, not the right one at all. Shit for now. New chat, the calendar.

Help me organize my day tomorrow. Tomorrow, I’m going to Shanghai. Warning. Preparing the user’s day is written.

Then I asked it What am I doing tomorrow?

Proposed planning:

8:30 productive light awakening.

I’ve been up since 7 for my school. I’d dream of waking up at 8:30 or 9 am.

Check all gentle hydration, 5 minutes of stretching, and open the Figma project.

What’s it talking about? It’s talking nonsense.

Not at all, because at 1 pm I have a train and at 1 pm I’m in deep work. You have access to my Google. Accomplish it based on that anyway, bro.

Okay, there it did it.

So awakening plus breakfast, stretching, quick email check at 8:30, 9:30. From 9:30 to 11:30, creative deep work. Finish script, prepare camera material, battery, SSD, and teleprompter.

I don’t know why.

Check your accessories and your shooting outfits.

That’s not bad.

At 11:30, quick lunch break. Plan a snack for the train. 12:15 12:50, pre-departure. Prepare your bag, tripod, microphone, and charger. Check tickets and reread the brief on Notion.

For now, it’s okay. Xi’an-Shanghai train.

Reading brief. Quick note for storyboard. Possibility of a micro nap if needed. Shooting location with address. Capture B-roll and A-roll. Check audio-video on site.

Indeed, it’s not at all me who said it, but you should indeed check a bit.

Quick file backup to SSD on the return train. Quick rush review on the return train. Partner email response if you have energy. Return to Xi’an, also a light dinner at 8 pm. Quick debrief. Check rush integrity. Note hot impressions from shooting. Free time or light editing at 9 pm.

What?

From 9 pm to 10:30 pm. Free time or light editing? Until 10:30 pm.

It’s making me work like crazy. Okay, it works. Well, I’ll tell you what it gives in the coming days. It doesn’t know I have classes and an internship. No, fortunately.

So, I’m telling you, this data scares me a bit more, and on rare occasions, it started thinking I had a child, so I went to remove it from his memory. Because of that, I find it a bit intrusive because it doesn’t only concern me, but it also concerns another person. So, I prefer it to be deleted.

Final Thoughts

Indeed, it doesn’t know that my evenings don’t look like that at all. The sender doesn’t want to be in ChatGPT. Already, if it can be active on my Gmail, it’s already over because all Gmails are read, analyzed for advertising, and the emails I receive also belong to me.

So, I don’t think it works like that. I think it does what he wants with my email.

---

So that was ChatGPT Pulse. What do you think, friends? Share in the comments, follow me, and subscribe to Nov Tech, of course.