Twelve years after pitching a concierge chatbot at TechCrunch Disrupt, Eugenia Kuyda has launched a product that finally matches that original ambition. Her startup Luka, which became Replika, was ahead of the deep-learning curve; now Kuyda's new company Wabi is releasing a 2.0 version of its personal AI assistant with a multithreaded messenger design.
Unlike single-thread assistants such as Instinct and Muse, Wabi 2.0 creates separate threads for different goals and automatically spins up apps from connected email—like a weight-loss tracker in a health thread or a kid-event finder in a family thread. The apps can be shared as group chats, so partners or family members can interact with the agent together, with scoped access to personal data like calendar or bank details.
Wabi's pivot reflects a finding that users liked personal software but didn't want to build it themselves. The app is free for now, and the company has not settled on a business model; Kuyda said Wabi is in talks to raise more funds. It is SOC 2 compliant and stores most user data in the U.S., though researchers caution that long agentic conversations can lose context and become less reliable over time.