Features

Your AI works inside NoteX.
NoteX works inside your AI.

Claude and ChatGPT run natively as add-ons, the web sidebar drafts and edits with your approval, and any agent reaches your knowledge through MCP. AI as a first-class citizen, not a bolt-on.

The architecture

Three ways in.
All of them native.

Add-ons on desktop

Claude Code, Codex, and Continue install from the curated catalog and run inside NoteX, with your notes as their working files. No copy-paste loop, no browser tab.

NoteX AI on web

A built-in AI sidebar that searches your workspace, reads pages, and proposes edits you apply with one click. Personas, prompt library, and a model picker included.

Any agent via MCP

A Notion-style REST API plus an MCP server give Claude, Cursor, or any agent scoped access to your knowledge base, with keys you control.

API & MCP

Desktop · add-ons

Claude works here.
Literally here.

Install Claude Code from the catalog and it opens inside NoteX with your notes folder as its workspace. Ask it to reorganize the wiki, draft a spec from meeting notes, or sweep every runbook for stale steps: it reads and edits the same files you do.

By design, no models ship in the box. You bring your own AI account, your conversations go straight to your provider, and NoteX stays a neutral platform. A dedicated agent sessions window and an AI customization system (agents, skills, prompts, instructions, hooks, MCP servers) round out the desktop story for people who run serious agent setups.

✳️
Claude CodeAnthropicAgentic AI on your notes as files
Installed
🤖
CodexOpenAIChatGPT inside your workspace
Install
🔄
ContinueContinueOpen-source assistant, bring any model
Install

Installed from the curated catalog. Your accounts, your providers, your files.

Web · NoteX AI

An assistant that proposes.
Never one that overwrites.

The web sidebar streams answers grounded in your actual workspace: it searches pages, reads content, and returns proposals as cards with a diff preview. Nothing touches a page until you click Apply, and Ctrl+Z undoes what you let through.

One-shot actions handle the everyday: summarize, improve writing, fix grammar, translate, change tone, extract action items. Select any text and Ask AI about just that. Ghost-text autocomplete finishes sentences when you want momentum. Personas and a shared prompt library keep the whole team's AI on the same voice, and threads can be shared by link with password and expiry.

Find every open action item across the wiki and group them by owner.

Searched 42 pages, found 9 open items in 4 notes. Proposing a new page:

Create page: Open Action Items9 items · grouped by owner · diff preview available
Proposed
Append link to Team WikiUnder Operating cadence
Proposed

Proposals, not surprises: review the diff, apply with a click.

The quiet advantage

Notes AI reads
for a fraction of the cost.

Every AI conversation is limited by context: how much your agent can read before it answers. Structured note files carry the same knowledge as a PDF at a fraction of the tokens, up to 50x smaller in our internal benchmarks. More context, faster answers, lower bills.

The full speed story
Scanned PDF
≈ 180K tokens
Word document
≈ 82K tokens
NoteX file
≈ 3.6K tokens

One 40-page spec, three formats, from our internal benchmarks. Illustrative.

Who this is for

If your team already lives in Claude or ChatGPT, NoteX makes those tools first-class citizens of your knowledge instead of visitors pasting fragments. If you are building agents, the API and MCP server make your notes programmable infrastructure. And if you just want better drafts, the sidebar is one click away. See how AI-powered teams run NoteX.

Frequently asked questions

Does NoteX include an AI model?

The desktop app deliberately ships none: you install add-ons (Claude Code, Codex, Continue) and bring your own provider account, so your data relationship is with your chosen provider. The web workspace includes NoteX AI with a picker across leading models, including Claude.

Can the AI edit my notes without asking?

No. On web, the sidebar returns proposals with diff previews; you apply or dismiss each one. On desktop, agentic add-ons like Claude Code show their changes as file edits you review, the same way developers review code changes.

What is MCP and why should I care?

The Model Context Protocol is an open standard that lets AI assistants use tools. NoteX ships an MCP server with ten tools (search, read, create, update, append, trash), so Claude Desktop, Claude Code, Cursor, or any MCP client can work your knowledge base with a scoped key.

Is my content used to train models?

No. NoteX never trains AI on your notes. Desktop add-ons talk directly to your provider under your account and their terms; the web AI processes only what you point it at.

What are personas and the prompt library?

Web features: personas shape how NoteX AI responds (tone, role, expertise), and the prompt library stores reusable prompts with variables that your whole workspace can share.

Which surface should AI-heavy teams start with?

Both have a role: the desktop app for agentic work on files (Claude Code), the web for collaborative drafting with the sidebar, and the API/MCP for automation. The AI teams use case walks through a full setup.

Your notes deserve a better engine.

A Notion-style block editor on the engine behind the world's most popular code editor. Notes stay files you own, extensions add what you need, and AI plugs in natively. Free to start, on web and desktop.

Free to start · Your files, your disk · Windows desktop & web