Text.ai Built Alfi, a Group Chat Bot That Actually Reads the Room

The Macro: The Group Chat Is the New Social Network

Something happened to social media that the platforms themselves are only beginning to acknowledge. People stopped posting. Not entirely, but the shift is real and measurable. The feed-based broadcast model, where you post to hundreds or thousands of followers, is losing ground to smaller, private group conversations.

The data supports this. Instagram’s most-used feature is DMs, not the feed. WhatsApp and iMessage group chats are where real coordination happens. Discord servers function as persistent group chats with channels. The public square is getting quieter while the back rooms get louder.

For Gen Z, this shift is even more pronounced. The generation that grew up with Instagram and TikTok is increasingly uncomfortable with permanent public posts. Finsta accounts, close friends stories, private group chats. The energy is moving toward smaller audiences and more ephemeral communication. Broadcast social is becoming a performance. Group chat is where people are actually themselves.

And yet the group chat experience has barely evolved since the early days of iMessage and WhatsApp. You text. You send memes. Someone shares a link. Someone else ignores the link. The group makes plans that fall apart. Someone sends a restaurant recommendation that nobody can find three days later when they actually need it. The tools for group communication have not kept up with the amount of life that now happens inside them.

A few companies have nibbled at this. Telegram added bots and channels. Discord added forums and threads. GroupMe still exists, mostly for college students. But nobody has built a group chat product that is natively intelligent, one that understands the context of conversations and can actually help the group function better.

That is the gap Alfi is targeting. Not another messaging app with a chatbot bolted on. A group chat where the AI is a participant with social awareness.

The Micro: Three Founders Who Have Seen Scale

Text.ai is a four-person team in San Francisco, part of YC’s Spring 2025 batch. The company was founded by Rushi Shah, Paras Maniar, and Prahar Patel.

The founding team is experienced in a way that matters for a consumer product. Rushi is the CEO, with time at Tesla and NASA. Paras is a co-founder with JPMorgan, McKinsey, and Gannett/USA Today on his resume. Prahar is the CTO, coming from Walmart, Eventbrite, OpenTable, and Amazon. These are people who have worked at companies that process millions of users and billions of transactions. They understand scale, and they understand consumer product dynamics.

The product, Alfi, is described as the first group chat with social skills. That sounds like marketing copy, but the feature set makes the concept concrete. Alfi lives inside your group chat and does things a helpful friend would do. It integrates restaurant booking through Yelp, so when someone says “where should we eat,” the group can actually book a table without leaving the conversation. It includes multiplayer image generation, so the group can create and riff on images together. There is a social calendar for coordinating plans. And a “Group Wrapped” feature that gives the group analytics on their conversations, similar to Spotify Wrapped but for your friend group.

The company started as text.ai, an AI assistant that worked across existing messaging platforms. SMS, iMessage, WhatsApp, Telegram. They processed over 10 million messages globally in that form. That is not a trivial number. It means they have real data on how people interact with AI in messaging contexts, what questions they ask, what tasks they try to accomplish, and where the experience breaks down.

The pivot from “AI assistant inside existing platforms” to “our own group chat app with built-in AI” is interesting and risky. On one hand, they are giving up distribution on platforms with billions of users. On the other hand, they are gaining control over the full experience. Building features like restaurant booking and social calendar would be impossible inside iMessage or WhatsApp, where third-party integrations are tightly restricted.

The positioning around Gen Z leaving broadcast social is deliberate and, I think, directionally correct. The question is whether the shift away from public social translates into willingness to adopt a new group chat app. Group chats are sticky precisely because the group is already there. Asking a friend group to move from iMessage to Alfi is a high-friction ask, no matter how good the product is.

The Verdict

I find myself genuinely uncertain about this one. The thesis is right. Group chats are underserved. The AI capabilities are real and differentiated. The team is strong. But the distribution challenge is massive.

Messaging apps have some of the strongest network effects in consumer tech. People do not switch because a new app has better features. They switch because their friends are already there. WhatsApp won in most of the world not because it was the best messaging app, but because it was the one everyone else was already using. iMessage dominates in the US for the same reason.

To break in, Alfi needs to find friend groups where the value of the AI features is high enough to overcome the friction of switching. College friend groups coordinating social plans might be the wedge. That is where the social calendar and restaurant booking features have the most obvious utility. If they can get a critical mass of friend groups on campus, the network effects work in their favor within that community.

The 10 million messages processed in the previous text.ai incarnation gives them useful data, but I would want to know how much of that usage was habitual versus one-time. Consumer AI assistants have a history of high trial rates and low retention.

At 30 days, I would want to see average group size and messages per group per day. If groups are small (3-4 people) and active (20+ messages daily), the product is working as a social tool. If groups are large and quiet, it is not. At 60 days, the question is whether Alfi-initiated actions (restaurant bookings, calendar events) are actually getting used or sitting idle. At 90 days, retention is everything. Consumer messaging apps either become daily habits or they die. There is very little middle ground.

The product is thoughtful. The team is capable. The market timing feels right. But messaging is one of the hardest categories in consumer tech to crack, and I have seen too many good products fail here to call this one with confidence.