Skip to main content
Replika founder Eugenia Kuyda smiles, holding a phone displaying the Wabi AI messenger app interface.

Editorial illustration for Replika founder Kuyda launches Wabi, multiplayer AI messenger app

Replika Founder Launches Wabi AI Messenger App

• 4 min read

Twelve years ago, Eugenia Kuyda walked onto the TechCrunch Disrupt stage and demoed IO, a chatbot built by her startup Luka that could book restaurants and plan vacations. That was 2013, long before transformer models and ChatGPT existed. Luka eventually morphed into Replika, the AI companion app that made Kuyda a known name in the industry and millions of dollars in revenue. She stepped down as Replika's CEO last year and started a new company, Wabi.

On Monday, Kuyda launched Wabi to the public: a multiplayer personal AI assistant built as a messenger app. Users can run several agentic conversation threads at once, each tied to a different goal, and build small custom apps inside them, a weight-loss tracker in a health thread, a local events finder in a family thread. The timing puts Wabi in direct competition with a crop of new personal AI assistants, including Muse and Instinct, all chasing the same promise of handling inboxes, flights, and daily logistics on a user's behalf.

The launch marks a shift from how Wabi first presented itself, though Kuyda frames it less as a pivot and more as the product catching up to what she originally pitched back in 2013.

On Monday, Kuyda officially launched Wabi for all users, a multiplayer personal AI assistant that functions as a messenger app, allowing users to maintain multiple agentic threads dedicated to different tasks and long-term goals. Within each thread, users can build and use custom apps, like a weight-loss tracker inside a health thread or an app showing nearby children’s events inside a family thread.

Why this matters

Kuyda's twelve-year arc from IO to Wabi is a useful marker for how far the underlying tech has moved, even if the product concept, an AI concierge that handles your life, has barely changed. That consistency is worth sitting with. If the same pitch that "worked" on a primitive chatbot in 2013 is now being rebuilt atop large language models, the interesting question isn't whether AI assistants are finally possible, it's why it took this long to get the interface right.

Wabi's bet is on structure: multiple persistent threads, custom mini-apps per task, a messenger format instead of a single chat window. For founders building in the assistant space, that's a signal worth watching, single-thread chatbots may be hitting a ceiling on usefulness. For researchers, Kuyda's own framing, that early deep-learning-free IO "still worked," is a reminder that user experience design can mask or substitute for model capability longer than we'd like to admit.

We'll be watching whether Wabi's multiplayer, multi-thread approach actually sticks with users, or just adds more friction dressed up as flexibility.

Common Questions Answered

What is Wabi and how does it differ from Replika?

Wabi is a multiplayer personal AI assistant messenger app launched by Eugenia Kuyda that allows users to maintain multiple agentic threads dedicated to different tasks and long-term goals. Unlike Replika, which focused on AI companionship, Wabi enables users to build and use custom apps within each thread, such as a weight-loss tracker in a health thread or an app showing nearby children's events in a family thread.

What was IO and how does it relate to Kuyda's current work with Wabi?

IO was a chatbot built by Eugenia Kuyda's startup Luka in 2013 that could book restaurants and plan vacations, demonstrated at TechCrunch Disrupt before transformer models and ChatGPT existed. Kuyda's journey from IO to Wabi demonstrates how the underlying AI technology has advanced significantly, though the core product concept of an AI concierge handling your life has remained relatively consistent over the twelve-year period.

Why did Eugenia Kuyda leave Replika to start Wabi?

Eugenia Kuyda stepped down as Replika's CEO last year to start her new company, Wabi, which represents an evolution of her vision for AI assistants. While the article doesn't explicitly detail her reasons, Wabi appears to be her attempt to build a more functional, task-oriented AI assistant platform that goes beyond Replika's AI companion model.

What does the article suggest about why AI assistants took so long to develop properly?

According to the article's conclusion, the interesting question isn't whether AI assistants are finally possible with modern large language models, but rather why it took this long to get the interface right. The fact that the same pitch from IO in 2013 is now being rebuilt on advanced LLMs suggests the challenge has been more about user interface and practical implementation than the underlying technology itself.

LIVE00:37Microsoft CEO Nadella warns AI agents are like insider threats