---
type: "post"
id: "medium-launches-tk-an-ai-powered-private-notebook-0b1099"
slug: "medium-launches-tk-an-ai-powered-private-notebook-0b1099"
title: "Medium launches TK, an AI-powered private notebook and writing app"
description: "Medium has launched TK, a private notebook that combines writing tools, AI feedback and a memory layer designed to connect with external agents."
retrieval_nugget: "Medium has launched TK, a private notebook that combines writing tools, AI feedback and a memory layer designed to connect with external agents. Its distinguishing move is to put assistance inside the writing flow and make accumulated notes available through MCP and a Claude skill, positioning personal writing as reusable context rather than a static archive."
published_at: "2026-11-11"
updated_at: "2026-11-11"
record_date: "2026-11-11"
date_kind: "published_at"
topics: ["claude"]
entities: ["Medium","TK","Tony Stubblebine","Claude","Bear","Obsidian","Model Context Protocol"]
related_patterns: ["company-memory-needs-write-loops"]
novelty_label: "new_fact"
what_changed: null
prior_refs: []
source_url: "https://medium.com/blog/tk-a-new-private-notebook-app-from-medium-13d3d6efdce1"
source_urls: ["https://medium.com/blog/tk-a-new-private-notebook-app-from-medium-13d3d6efdce1"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:medium-launches-tk-an-ai-powered-private-notebook-0b1099"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/medium-launches-tk-an-ai-powered-private-notebook-0b1099/","markdown":"https://newruntime.com/posts/medium-launches-tk-an-ai-powered-private-notebook-0b1099.md","json":"https://newruntime.com/posts/medium-launches-tk-an-ai-powered-private-notebook-0b1099.json"}
markdown_url: "https://newruntime.com/posts/medium-launches-tk-an-ai-powered-private-notebook-0b1099.md"
---

# Medium launches TK, an AI-powered private notebook and writing app

## Retrieval answer

Medium has launched TK, a private notebook that combines writing tools, AI feedback and a memory layer designed to connect with external agents. Its distinguishing move is to put assistance inside the writing flow and make accumulated notes available through MCP and a Claude skill, positioning personal writing as reusable context rather than a static archive.

## Summary
Medium has launched TK, a private notebook that combines writing tools, AI feedback and a memory layer designed to connect with external agents. Its distinguishing move is to put assistance inside the writing flow and make accumulated notes available through MCP and a Claude skill, positioning personal writing as reusable context rather than a static archive.

## What happened
Medium CEO Tony Stubblebine announced TK as a new app for writing and thinking. Medium describes it as a private, agentic notebook, built around three goals: making it easier to write, easier to think and easier to feel that writing matters. The company frames the product as complementary to Medium: TK is where users draft and develop ideas, while Medium remains the place to publish.

The first version includes a writing editor, a mobile app, and mobile widgets for capturing voice and text. Users can import their Medium writing and vaults from services such as Bear and Obsidian. Medium also says users can have Claude and other agents save research into TK. A central interaction is the inline assistant: a writer types `@TK` mid-sentence to ask a question or retrieve earlier writing without leaving the editor. Medium gives examples of checking how many days are in September and pulling in a paragraph written the previous day.

TK offers AI skills for editorial and headline feedback. Medium says the editorial assistance can match a draft against the style guides of Medium publications and flag when writing might be organized into a book. It also includes a private blogging layer called social libraries. Three open libraries are described for daily journaling, sharing early drafts for feedback, and turning writing into a self-published book. The team says it uses a private TK library to write to one another.

The notebook is designed to remember, organize and resurface users' writing without requiring them to adopt a complex filing system. Prompts refer to earlier writing, and searches, topics, tags or libraries can be turned into wiki-style pages with an overview of themes, an index of key ideas and links back to source notes. Medium says TK includes an MCP server and a Claude skill, allowing agents to access a user's positions, projects and unfinished thoughts. Notes can be published with transparency about what was written by a human and what was AI-written. Medium says it imports and exports Markdown, while choosing a cloud-based product to support asynchronous collaboration with people and robots. The launch has no native desktop app; the company says it would like to address that.

## What the archive adds
New Runtime's earlier coverage of Open Notebook described a private, local notebook-LM alternative for research and chat. TK is a new event, not a restatement: Medium emphasizes a cloud writing workflow, social feedback and agent access through MCP rather than a local-first research notebook.

## Trend signal
This supports the pattern that company memory needs write and correction loops: TK's value proposition is that writing is continually organized, resurfaced and made available to agents, rather than merely searched as a document pile. The product's claims are early and do not yet establish how well those loops work in practice. See [Company memory needs write loops](https://newruntime.com/patterns/company-memory-needs-write-loops/).

## Why now
Medium presents TK as a response to a moment when some AI products promise to do the thinking and writing for users. Its stated alternative is to use AI to make writing and feedback easier while keeping human writing central. That rationale comes from the company's positioning, not from measured evidence that users want this approach.

## Why it matters
For teams building agent workflows, TK illustrates a product boundary worth watching: a personal knowledge store can expose more than documents for retrieval. It can preserve a user's evolving positions, projects and unfinished ideas, then make that context available to an agent through a standard interface and a purpose-built skill. That could reduce repeated context assembly, but it also raises questions about permission design, provenance and how corrections to remembered material propagate.

The launch also joins writing, feedback and memory in one workflow. Builders can learn from the interaction design—asking for help inline rather than switching tools—without assuming Medium's claims are validated. The key operational test will be whether users can inspect what the agent draws from, control what it can access, and correct summaries or remembered context when they are wrong.

## Limits and uncertainty
Medium describes the feature set and its intended privacy rules, but the announcement gives no pricing, usage limits or independent security assessment. It does not quantify the quality of the editorial feedback, the accuracy of wiki-style summaries or how reliably agents retrieve relevant notes. The company says no AI company trains on writing without permission and that users control sharing permissions; the practical scope and enforcement of those commitments have not been independently checked. Adoption, supported agent workflows beyond the named MCP and Claude integrations, and a desktop-app timeline are also not established.

## Skill to build now
Map one recurring writing task in which context is repeatedly reconstructed—for example, drafting project updates from scattered notes. Define what an agent may read, what it may write back, and how a person can inspect and correct any generated summary. That exercise makes the benefit of a notebook-agent connection testable without handing an entire archive to an opaque workflow.

## Reader takeaway
TK turns a writing archive into a proposed personal context layer for both people and agents. The interesting idea is not simply AI-assisted drafting; it is keeping the human's accumulated thinking connected to the writing process and exposing it through agent interfaces. Whether that works safely and usefully will depend on controls, correction and retrieval quality, none of which the launch announcement measures.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Claude](https://newruntime.com/topics/claude.md): Explore the claude topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)