Replika’s founder built the AI assistant she pitched at Disrupt 12 years ago

Your AI assistant can now manage separate conversations and tasks without confusing them, and share that context with others.

Eugenia Kuyda, founder of the AI companion chatbot Replika, has launched a new personal AI assistant called Wabi. This launch fulfills a vision Kuyda first presented with her earlier chatbot, IO, at TechCrunch Disrupt 12 years ago. Wabi is designed as a multiplayer personal AI assistant that functions within a messenger app. It allows users to maintain multiple "agentic threads" for different tasks and goals, each capable of running custom applications, such as a health tracker or event finder. This new iteration of Wabi, also known as Wabi 2.0, evolved from its initial concept as a social platform for mini-app creation. The current version automatically generates threads and apps upon connecting to user data like email, with users able to request new custom apps or threads. Wabi aims to differentiate itself from competitors like Instinct and Muse by offering this multi-threaded, multiplayer functionality, addressing the context loss and hallucination issues that can arise in single-thread AI conversations. Kuyda is also exploring shared threads to facilitate collaboration between users and their AIs.

Wabi's multiplayer and multi-threaded design addresses a critical limitation in current AI assistants: context decay and information overload. By segmenting conversations and tasks into distinct threads, Wabi aims to provide a more organized and reliable user experience, preventing the degradation of AI performance over time that occurs in single, continuous chat interfaces. This approach also enables collaborative AI experiences, allowing multiple users to interact with AI agents and share information within specific, task-oriented contexts.

The evolution of Wabi from a user-driven app-building platform to an agent-driven one signifies a market shift towards more intuitive and less labor-intensive AI personalization. As the personal AI assistant market becomes increasingly crowded with offerings like Muse and Instinct, Wabi's strategy of having the AI autonomously generate and manage applications within threads presents a compelling alternative. This moves away from users needing technical skills to build their own AI tools, making advanced AI functionality more accessible and functional for a broader audience.