From Early Chatbots to Advanced AI Assistants: The Journey of Eugenia Kuyda and Wabi
Twelve years ago today, Eugenia Kuyda took the stage at TechCrunch Disrupt to introduce IO, a conversational chatbot that functioned as a lifestyle concierge. Designed to assist users in booking restaurants, planning vacations, and managing other daily tasks, IO was the inaugural product of Kuyda’s startup Luka. This pioneering technology emerged well before the deep learning breakthroughs that underpin today’s large language models.
Reflecting on IO’s early capabilities, Kuyda told TechCrunch, “It still worked. It was pretty magical… but, of course, it’s nowhere near what we have today.” Despite its limitations by today’s standards, IO laid the foundation for what would evolve into more sophisticated AI-driven personal assistants.
The Evolution from Luka to Replika and the Birth of Wabi
Luka eventually transitioned into Replika, a widely recognized AI chatbot companion platform. Replika’s mission centered on creating AI that could foster emotional connections and help users combat loneliness. After stepping down as Replika’s CEO last year, Kuyda launched a new venture, Wabi, aiming to realize her long-held vision of personal AI assistants with enhanced versatility and social integration.
Wabi enters the market amid a burgeoning wave of AI assistants like Muse and Instinct, both promising to manage tasks from email inboxes to travel planning. However, Wabi differentiates itself through its multiplayer and multithreaded design, allowing users to engage multiple AI-driven conversations simultaneously, each dedicated to specific goals or topics.
Introducing Wabi 2.0: A Multiplayer Personal AI Assistant
On Monday, Kuyda officially launched Wabi for all users. Unlike traditional AI assistants that operate within a single conversational thread, Wabi allows users to maintain multiple agentic threads, each functioning as a dedicated space for different tasks or long-term objectives. For example, users can have a health thread with a custom weight-loss tracker app or a family thread featuring an app that displays nearby children’s events.
This launch represents an evolution rather than a pivot. Initially, Wabi was introduced as a social platform where users could create and share mini apps with friends or across the community. After raising $20 million in a pre-seed funding round last year, Kuyda and her team realized that while users favored personalized software over generic app store solutions, many preferred not to spend time building these apps themselves.
With Wabi 2.0, the AI agent takes the lead in app creation. Upon connecting Wabi to a user’s email, it automatically generates relevant threads and apps to get started. Users can also request custom apps or initiate new threads, enhancing flexibility and personalization.
“It’s doing what Instinct and Muse do, but it also has all these apps, which helps users do more stuff,” Kuyda explained, highlighting Wabi’s unique blend of capabilities.
Multithreaded and Multiplayer: Redefining AI Interaction
One of Wabi’s standout features is its multithreaded, multiplayer design. Unlike single-threaded AI assistants such as Instinct and Muse, which can overwhelm users with a continuous stream of information, Wabi organizes conversations into distinct threads. Kuyda points out the drawbacks of single-thread experiences: “They get worse over time because it’s just so much context in one endless conversation.”
AI researchers have noted that prolonged chatbot conversations increase the risk of hallucinations, context loss, or inappropriate responses, even when safety guardrails are in place. Wabi’s threading approach helps mitigate these issues by compartmentalizing interactions.
Beyond individual threads, Wabi supports shared threads that operate like group chats. Kuyda envisions this multiplayer aspect as a solution to the current norm where “all the agentic apps are just [conversations] between you and the agent.” She believes AI should facilitate collaborative experiences, enabling groups of people to interact together with their AI assistants.
Demonstrating this, Kuyda shared examples of her own shared threads. One thread with her boyfriend includes apps for live music tracking, daily relationship questions, restaurant recommendations, shared calendar events, and book lists with reading suggestions. In another family thread, she uses apps to monitor her daughter’s schedule, find kid-friendly events, and quiz her daughter on Japanese vocabulary.
Users can control the level of information shared in these threads, from emails and calendars to even bank account details. “So my partner can ask Wabi in the chat, ‘Is Eugenia free next Thursday?’ or ‘Did she buy those tickets?’” Kuyda noted, emphasizing that access permissions are scoped to each chat and visible to all participants for transparency.
Security, Privacy, and the Human Element of AI
Privacy remains a cornerstone of Wabi’s design. The platform is SOC 2 compliant, indicating that its data security controls have undergone independent audits. Wabi operates entirely in the cloud, with the majority of user data stored within the United States.
Kuyda reflects on her journey, stating, “My previous company Replika worked on AI that will make people feel… less lonely. To a degree, that is [in Wabi] as well.” She highlights how AI can foster stronger human connections by helping people manage their time and activities together — whether that means attending concerts, reading books, or simply spending more quality time.
Currently, Wabi is free to use. The company is still exploring viable business models and is in ongoing discussions to raise additional funding to support its growth and development.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Source: Here
