Messaging Opportunity and Group Chat Dynamics
OpenAI has started piloting group chats for ChatGPT, initially in Japan, New Zealand, South Korea, and Taiwan. It’s a simple idea with potentially big implications: bring multiple people and the model into the same conversation so the AI can search, summarize, draft, compare options, and keep a running context for everyone at once.
This is OpenAI’s another clear attempt to evolve from a single‑player Q&A box to try to graduate to a social surface. Session‑based usage has a ceiling; chats with friends, family, or teams can basically run all day. Making ChatGPT the “nth participant” in a thread can potentially fill the engagement gap. As Scuttleblurb mentioned in our recent podcast (Spotify, Apple, YouTube, RSS feed), there are some apps that require effort to use, and there are some that require conscious effort to NOT use! The reality is despite ChatGPT’s incredible capabilities, the median user likely requires too much agency to stay engaged as DAU over the long term. A persistent stream of notifications from multiple group chats on ChatGPT will shift ChatGPT from the latter to much closer to the former over time.
Perhaps very few people appreciate more how messaging has become the base layer of social interaction these days than the CEO of the company owning two largest messaging apps in the world: Mark Zuckerberg. In an interview with Stratechery in May 2025, Zuckerberg laid out how our social interaction evolved over time:
It used to be that you interacted with the people that you were connecting with in feed, like someone would post something and you’d comment in line and that would be your interaction.
Today, we think about Facebook and Instagram and Threads, and I guess now, the Meta AI app too and a bunch of other things that we’re doing, as these discovery engines. Most of the interaction is not happening in feed. What’s happening is the app is like this discovery engine algorithm for showing you interesting stuff and then, the real social interaction comes from you finding something interesting and putting it in a group chat with friends or a one-on-one chat. So there’s this flywheel between messaging which has become where actually all the real, deep, nuanced social interaction is online and the feed apps, which I think have increasingly just become these discovery engines.
Given this context, I have seen some people expressing their surprise online that Meta hasn’t run after this opportunity before OpenAI did. Well, Meta did.
WhatsApp does let you mention Meta AI in individual and group chats, but end‑to‑end encryption means the AI only sees what you explicitly send it or what you opt to summarize; by design, it can’t freely graze the entire conversation. You cannot wish for end-to-end encrypted messaging and the AI being a unprompted participant. And even when you explicitly ask for AI’s help, it may not perform as well as you would like due to their lack of context of the entire conversation.
That’s not going to be a problem for ChatGPT. ChatGPT will have complete context, allowing it to function as a fully informed, dynamic “nth participant.” More importantly, users starting a group thread on ChatGPT are doing so clearly to utilize the AI, mitigating any immediate privacy concerns that plague established messaging apps trying to retrofit AI into a privacy-first environment.
I suspect it is going to be very, very difficult for Meta to change their tune on end-to-end encryption. It’s not only technical constraint, rather perhaps this is yet another “strategy tax” for Meta for having a dubious privacy perception from its users. As a result, I echo with Ben Thompson that it will be lot harder to borrow the page from OpenAI here even if it becomes quite popular.
However, such analysis underestimates the layers of moats of the current messaging apps. Take WhatsApp, for example. If you are reading this sitting in the US or perhaps much of the western hemisphere, it may be fundamentally difficult for you to appreciate just how deeply reliant some countries are on apps such as WhatsApp. Dharmesh Ba last month wrote an intriguing piece “WhatsApp owns India!” which might help you understand this dynamic:
“In 2023, I met a cosmetic store owner in Nagpur who had accidentally invented demand forecasting.
He sells beauty products - Korean serums, Indian sunscreens, foreign moisturizers. His challenge wasn’t competition or pricing. It was more existential: his shelf space could hold maybe 200 SKUs, but his customers now wanted access to 2,000.
Ten years ago, this wasn’t a problem. A woman would walk into his store, and he’d guide her. “Try this Lakmé sunscreen. Let me show you how to apply it.” He hired young women specifically for this - people who could explain products, who could teach.
But Instagram changed the game. Women no longer came to his store to discover products. They came with screenshots, asking for specific Korean brands they’d seen on reels. If he didn’t have it, they left. And he couldn’t stock everything.
So he invented a system.
He joined a vendor WhatsApp group - wholesalers who posted about new products arriving in the market. When something looked promising, he’d post the image to his WhatsApp status. Or he’d share it in a customer group he’d built over years.
Then he’d wait. If even two or three people asked for the price, he knew there was demand. He’d place an order before the product even touched his shelf.
He wasn’t using WhatsApp for communication and marketing alone but turned it into a just-in-time inventory system.
This was India’s answer to the Shopify + Stripe stack. Discovery happened on Instagram. Demand validation happened on WhatsApp Status. Payment happened on UPI. Small businesses had discovered that consumer apps could be hacked into business tools. In this WhatsApp owns India’s communications layer. And unlike every other tech platform, there’s no competitor even close. Instagram competes with Twitter and YouTube. WhatsApp has... nothing.”
Of course, WhatsApp got to this status by aggregating the users first. Moving everyone to a new group chat sounds very simple at first, but let me tell you my personal experience. I am in several DM groups on twitter (or X). If you use twitter DM, I don’t need to explain to you what a sub-par experience the whole thing is compared to any other messaging apps. What used to really bother me is I couldn’t search anything in my conversation (we can now), a basic feature that’s been available on WhatsApp or Messenger for forever. So, I once took the initiative to move couple of my twitter DM groups to WhatsApp. Well, half the people became completely inactive there probably because they barely open the app. So, instead of solving any problem, I became more confused where to share my thoughts and eventually had to come back to twitter. Just last week, Twitter was able to somehow exacerbate the messaging experience even more, but no matter what Twitter does, I think it is more likely than not that my DM groups will remain stuck on Twitter. If we have such hard time to move off twitter, I am fairly comfortable thinking it will be even harder to move away from messaging apps that are much more functional and useful.
Nonetheless, I do think there will be some chats that are obviously a better fit for ChatGPT than WhatsApp/Messenger. Imagine going for a trip to Europe with your family or/and friends. Instead of everyone individually making plans by searching things on Google, everyone going to the trip can just simultaneously plan on ChatGPT. I can totally see how that can be materially a better experience than what the current messaging apps can deliver today.
However, while ChatGPT’s hands are not tied due to privacy, they do have different set of limitations. While everyone can join your group chats on ChatGPT, ChatGPT’s responses to your queries can depend on your subscription tier. If someone in the group chat has Pro tier and hence has access to the most advanced models compared to a free user who can be bound not only by lower quality models but also rate limits, that can create a bifurcated experience even within the group chat, diluting the overall experience for all participants. Even if OpenAI eventually launches ads and addresses these limitations by offering more uniform group chat experience, people may become less comfortable with the idea that ChatGPT is showing them ads based on the private conversations they are having with their friends. Perhaps the revealed preference of going to ChatGPT to open a group chat itself will be strong enough to subdue such “AI is reading my conversation” concerns, but it can still be tricky to navigate the privacy quagmire.
While much of the focus of this launch seemed to be focused on consumer use cases, I believe where this gets strategically much more interesting is at work. In an enterprise, the company can just standardize the plan, eliminating the “fairness” problem, and the group can collaborate on the same thread with predictable capability. That looks a lot less like “threat to WhatsApp” and perhaps eventually more of a problem for Slack and Teams.
Of course, Slack and Teams aren’t asleep. Slack AI already summarizes channels and threads, produces recaps, and is layering in richer context features. Microsoft’s Copilot can be added to group chats and channels and will summarize discussions, answer questions, and ground responses in the prompter’s permissions.
As with many things in AI these days, the competitive dynamic seems quite fluid. OpenAI itself likely acknowledges this which is why they keep approaching the opportunity set in front of them in a much more open ended fashion than a narrow focus on a specific market. Of course, once they can see what the users are mostly using group chats for, they can always focus on building more features to cater to such use cases later. If ChatGPT “group chats” proves to be mostly people collaborating for work, WhatsApp may be wrong place to wonder how that will evolve. But if it’s more about couples doing joint therapy sessions with ChatGPT, it may pose a threat to an entirely different market altogether!
In addition to “Daily Dose” (yes, DAILY) like this, MBI Deep Dives publishes one Deep Dive on a publicly listed company every month. You can find all the 64 Deep Dives here.
Current Portfolio:
Please note that these are NOT my recommendation to buy/sell these securities, but just disclosure from my end so that you can assess potential biases that I may have because of my own personal portfolio holdings. Always consider my write-up my personal investing journal and never forget my objectives, risk tolerance, and constraints may have no resemblance to yours.
My current portfolio is disclosed below: