TabX positions itself as the memory and organization layer that simple page assistants lack. Instead of answering questions only about the page currently on screen, it keeps a private, per-user index of everything you have browsed, so a vague description like "the pricing breakdown I read three weeks ago" can surface the exact page again. The side panel opens with Ctrl+I (Cmd+I on Mac) directly on any site.
Beyond recall, TabX organizes your working session. Open tabs automatically cluster into named, color-coded Chrome groups by intent, and your visits are auto-assembled into named research 'journey' timelines that can be reopened in the original browsing order with one click. You can @mention any tab, stack, or timeline to ask a question and get answers grounded in, and cited back to, the pages you actually read. It also offers an ad-free AI web search that returns three to five hand-ranked links with a short answer and suggested reading order, then opens results scrolled to the exact passage.
The team emphasizes privacy: content is indexed into your own private cloud memory, de-duplicated, never captures password fields, and is never sold or shared. You can request a memory wipe by email. TabX works notably well on heavy web apps such as Gmail, GitHub, ChatGPT, WhatsApp Web, and YouTube.
