Wabi 2.0 Launches Multi‑Threaded, Multiplayer Personal AI Assistant

Twelve years after pitching a lifestyle concierge chatbot at TechCrunch Disrupt, Eugenia Kuyda has turned that early vision into a shipping product. Her new startup, Wabi, moved from a social mini‑app platform to a full‑featured personal assistant that organizes work, health, travel and family tasks in separate, persistent threads.

Wabi 2.0 automatically spins up relevant threads the moment a user connects email or calendar data. Inside each thread the agent builds custom micro‑apps — such as a weight‑loss tracker in a health thread or a local‑events board in a family thread — without any manual coding.

Unlike single‑thread rivals such as Muse and Instinct, Wabi keeps contexts isolated, reducing hallucination risk and preventing the endless‑scroll fatigue that plagues long conversations. Threads can also be shared, turning an individual assistant into a collaborative group chat.

Why it matters for GPU / AI infrastructure

Multi‑threaded, always‑on personal agents generate continuous inference demand across many concurrent sessions. Scaling this workload economically requires elastic GPU clusters with low‑latency interconnects — exactly the capacity profile that aigpu cloud platforms are built to deliver.