Back home

TK

web 0.91
Visit source

Intro

TK is Medium's private notebook / writing app, introduced 9 November 2026 by CEO Tony Stubblebine. The name comes from the journalist's placeholder “To Come”: type @TK mid-sentence to call an assistant without leaving the keyboard. Anyone with a Medium account can join.

It is not a generic note organizer — it aims to make writing the hub of your digital life: think, remember, then influence. v1 includes a focused writing surface, mobile app and widgets (voice/text), instant import of everything you wrote on Medium, and imports from Bear / Obsidian vaults; agents can stash research into TK too. Feedback comes from AI (editorial/headline, Medium publication style guides, “could this be a book?”) and humans via social libraries (Public Journal, Get Feedback on Your Writing, Write a Book). Any search, topic, tag or library can summarize into a wiki-style overview. An MCP server and Claude skill let agents use your positions, projects and half-finished thoughts as personal context.

Highlights:

  • Announcement: https://medium.com/blog/tk-a-new-private-notebook-app-from-medium-13d3d6efdce1 ;
  • “TK is the gym, Medium is the stage” — practice vs public performance;
  • Values: no training on your writing without permission, optional human/AI provenance, Markdown import/export with cloud collaboration as the primary design;
  • No native desktop app at launch (acknowledged on the roadmap);
  • For people who write to think and want personal context for agents — not a slop factory.

Explain

Why it matters

What agents lack is often personal intelligence, not general knowledge. TK wires a private writing corpus to MCP / Claude skill instead of pasting Markdown by hand. Inline @TK and social libraries try to keep focus while still getting cheap feedback; Medium's profitability claim (since August 2024 in the post) lowers “will this app vanish” risk.

Limits

Cloud collaboration wins over pure local Markdown; missing native desktop hurts keyboard-heavy workflows. Library and AI feedback quality depend on community and models. Complex Obsidian habits may not map 1:1 — TK deliberately avoids forcing a heavy organization system.

Usage

  1. Read the Medium blog post and join with your Medium account;
  2. Import Medium / Bear / Obsidian writing; try @TK mid-sentence for facts or yesterday's paragraph;
  3. Use social libraries for feedback; attach MCP / Claude skill when agents need your context;
  4. Publish on Medium when ready; export Markdown for backup as needed.

Advertisement