Replika Founder Eugenia Kuyda Launches Multiplayer AI Assistant Wabi

Eugenia Kuyda, the entrepreneurial technologist who founded viral chatbot companion platform Replika, officially launched her new personal artificial intelligence software platform Wabi to the public on Monday. The public release realizes the core personal assistant architecture that Kuyda originally pitched twelve years ago on stage at the TechCrunch Disrupt conference with early venture Luka. Wabi functions as a multi-threaded messaging application designed to support collaborative multiplayer interactions between groups of friends, family members, and autonomous software agents. Kuyda stepped down as chief executive officer of Replika last year to establish Wabi as an independent artificial intelligence venture.

Evolution from Early Chatbots to Autonomous Agents

In 2014, Kuyda debuted conversational software prototype IO under her initial startup Luka prior to the arrival of modern deep learning transformer architectures. While Luka subsequently pivoted into emotional companion Replika, Kuyda affirmed that recent frontier model breakthroughs finally allow real-time autonomous task execution. The commercial launch marks an evolutionary progression for Wabi from custom micro-app generation toward a unified conversational agent ecosystem. Kuyda explained that the platform combines ambient conversation with dynamic interactive mini-applications to empower everyday users to accomplish substantive activities.

Multi-Threaded and Multiplayer Group Architecture

Diverging from traditional single-thread chatbots like Instinct and Muse, Wabi organizes user workflows into segmented conversational tracks. Kuyda noted that single-thread chatbots inevitably bury crucial information, whereas multi-threaded channels preserve context across extended conversations. Kuyda criticized conventional AI agent applications for restricting digital interactions strictly to isolated one-on-one dialogues between an individual user and a model. Wabi enables collaborative shared threads where friends and domestic partners can simultaneously query the assistant regarding calendar availability or event tickets. Shared threads display interactive software widgets across conversation headers, allowing participants to jointly track grocery deliveries, travel reservations, and shared itineraries.

Data Security Protocols and Privacy Frameworks

To address consumer privacy concerns surrounding conversational data, Wabi completed independent audits achieving SOC 2 security compliance certification. The technical architecture ensures that shared group threads only access calendar information and personal data explicitly authorized by participating individuals. Kuyda reflected that while Replika concentrated on alleviating individual loneliness, Wabi focuses on fostering human relationships and shared real-world connections.

Commercial Availability and Venture Trajectory

Industry observers noted that underlying computing technology has finally caught up with Kuyda's original product vision just as personal AI assistants gain mainstream traction. Wabi is currently available free of charge across mobile platforms while executive leadership explores future monetization strategies. The startup is actively conducting discussions with venture capital investment firms to secure additional external expansion funding.

How Wabi differs from a standard chatbot

Wabi is presented as a messenger-like space for multiple AI-assisted conversations or projects rather than one continuous assistant thread. Users can build task-specific mini-apps within those conversations, and some features are intended to be shared or used collaboratively. That is a product-design distinction, not proof that the system can complete arbitrary tasks on a user's behalf without errors. Kuyda's earlier work on Replika supplies background, but Wabi should not be described as the same company or simply an upgrade of the old companion.

Everyday usefulness and risks

For practical tasks such as travel planning, family organisation or document handling, an AI-generated result is only as good as the tools and permissions behind it. A timetable or booking suggestion is not a confirmed reservation, and a generated health tracker should not be mistaken for professional medical advice. Shared agent threads also create questions about who can see a conversation, grant access to accounts or revoke permissions. The October 19 public launch makes the service available for testing; evidence of reliability, retention and sustainable pricing will emerge only through actual use and transparent product documentation.