Unabyss

Unabyss is a shared context layer for people who use more than one AI tool. It connects to the systems where your work actually lives, builds a single portable memory from them, and exposes that memory to every assistant over the Model Context Protocol — so the background you gave Claude this morning is already there when you open Cursor, ChatGPT, or Perplexity this afternoon.

The problem it targets

Each assistant keeps its own island of memory. Your code editor does not know what was decided in Notion, your research tool has never seen the Linear ticket, and every new session starts with the same paragraph of re-explanation. Unabyss treats that re-explanation as the bug and removes it by making one context set readable by all of them.

What it connects

Sources span productivity and knowledge tools (Notion, Obsidian, OneNote, Google Drive, Linear), communication (Gmail, Slack, GitHub, LinkedIn, X), calendars, and meeting recorders including Fathom, Fireflies, tl;dv, and Granola. On the consumer side it plugs into Claude, ChatGPT, Gemini, Codex, Cursor, VS Code, OpenCode, and Antigravity. Because delivery is MCP-native rather than a per-app plugin, adding a new client is a configuration step rather than an integration project.

Design stance

Two ideas separate it from a chat assistant's built-in memory. The context is self-updating: it is derived from live sources instead of being a static file you maintain by hand. And it is yours — a context you own and point at whichever model you happen to be using, rather than a vendor-locked profile that disappears when you switch tools.

Unabyss is free to start with paid plans above that, has synced over 650,000 items to date, and was named Product of the Day, Week, and Month on Product Hunt.