Anytype: A Local-First, P2P Notion Alternative, and How to Export Its Web Research to Markdown

Anytype is one of the more interesting Notion alternatives because it starts from a different assumption: your workspace should work locally first, then sync peer to peer when you want it to. That makes it appealing if you like Notion's flexible databases and pages, but you do not want every note, draft, or research clipping to live only in a cloud account.

I tested Anytype as a research workspace for a small writing project. I used it to collect sources, outline sections, and connect notes about tools in the local-first productivity space. I also tested Web2MD, our Chrome extension, as the bridge between web pages and AI tools like ChatGPT, Claude, and Cursor.

The short version: Anytype is strong when you want a private, structured knowledge base. Web2MD is useful when you need to bring messy web pages into that knowledge base, or into an AI chat, as clean Markdown.

What makes Anytype different

Anytype describes itself as local-first and peer to peer. In practical terms, that means your data is stored on your device, and the app is designed to keep working even when you are offline. Sync is not the same model as a traditional web app where the server is the source of truth.

That matters for personal knowledge management. A lot of people use Notion for everything: meeting notes, journal entries, project specs, reading lists, and sometimes sensitive client research. Notion is polished and collaborative, but it is still a hosted service. If your use case is private notes first and collaboration second, Anytype's architecture feels more aligned.

In my testing, the main mental shift was objects instead of plain pages. Anytype lets you create notes, tasks, bookmarks, projects, people, and custom object types. Those objects can have relations, which behave a bit like database properties in Notion. It is flexible, but it has a learning curve. If you want a blank page and a table, Notion is easier on day one. If you want a personal system that can become a graph of connected things, Anytype is worth the effort.

Where web research gets messy

The hard part is not taking notes inside Anytype. The hard part is getting the web into a usable format.

Most articles, docs, and product pages are not written for AI tools or note apps. They include nav bars, cookie banners, related posts, tracking scripts, sidebar links, and repeated footer content. Copying and pasting into Anytype or an AI chat often brings along too much clutter.

For my test, I used Web2MD to convert pages about Anytype, local-first software, and note-taking workflows into Markdown. The extension runs inside Chrome, so it reads the page I am already viewing and converts the visible document structure into Markdown. The result is much easier to review, summarize, quote, or paste into Anytype.

Here is a simplified example of the kind of Markdown output I used for a product research note:

# Anytype

Anytype is a local-first workspace for notes, tasks, databases, and connected objects.

## Key ideas

- Local-first storage
- Peer-to-peer sync
- Object-based knowledge management
- Offline access
- End-to-end encrypted backup and sync options

## Notes from testing

Anytype feels closest to Notion when building dashboards and collections,
but closer to a personal knowledge graph when linking objects together.

That is the shape I want before sending content to an AI model. Headings are clear. Bullets stay bullets. The page title is preserved. The noise is mostly gone.

Why browser-side conversion matters

There are good web-to-Markdown tools already. Jina Reader is fast and simple for public URLs. Firecrawl is powerful for developers who need crawling, extraction, and API workflows. MarkDownload is a useful open-source browser extension for saving pages as Markdown.

I do not think those tools are bad. They solve real problems.

The reason Web2MD exists is a narrower problem: sometimes the page you need is already open in your browser, behind a login, in a private workspace, or on a paywalled site you personally have access to. A server-side reader cannot fetch that page because it does not have your session. Even when it can fetch the public version, it may not see the same content you see.

Web2MD converts the page in your browser. That gives it three practical advantages:

• It works on logged-in pages that public URL fetchers cannot access
• The content does not need to be sent to a third-party extraction server
• You can inspect the page first, then convert exactly what you meant to convert

This is especially relevant for Anytype users, because the same people interested in local-first notes often care about privacy and control. If you are collecting research from a private dashboard, a members-only article, or internal documentation, browser-side conversion is a better fit than pasting a URL into an external service.

Using Web2MD with Anytype

My workflow was simple.

First, I opened the source page in Chrome. Then I clicked Web2MD and converted the page to Markdown. Before copying it, I checked the built-in token counter. That is a small feature, but it matters if the next step is ChatGPT, Claude, or Cursor. You can see whether the content is small enough for your prompt, or whether you should trim it first.

Then I either copied the Markdown into Anytype as a source note or used one-click send-to-AI to ask for a summary, critique, or outline.

A research clipping ended up looking like this:

# Local-first software

## Summary

Local-first software keeps the primary copy of user data on the user's device.
Network sync is used for collaboration and backup, not as the only place where
the data lives.

## Why it matters for note-taking

- Notes remain available offline
- Users are less dependent on a single SaaS provider
- Sync can be designed around privacy
- Backups and exports are easier to reason about

## Questions to explore

- How does conflict resolution work across devices?
- What happens if a sync node disappears?
- Can non-technical users understand the model?

From there, I could create an Anytype object for the source, relate it to a larger project, and keep my own notes separate from the extracted material.

Honest limits

Web2MD is not trying to be a full crawler. If you need to crawl hundreds of pages, schedule extraction jobs, or build a data pipeline, Firecrawl is probably the better fit. If you only need a quick Markdown version of a public article, Jina Reader is excellent. If you want a highly configurable open-source browser extension, MarkDownload is worth a look.

Web2MD is focused on a specific AI workflow: open a page, convert it locally, check token size, and send or copy clean Markdown.

There are also product limits to know. Web2MD is Chrome-only today. The free tier includes 3 conversions per day. Pro is $9 per month if you need more. That is not the right pricing model for everyone, but it keeps the basic version usable without an API key and without making you set up a developer account.

Anytype has limits too. It is not a one-to-one Notion clone. Some workflows that are obvious in Notion take more setup in Anytype. Collaboration is not as instantly familiar as sharing a Notion page with a team. The object model is powerful, but you have to learn it.

Where Anytype and Web2MD fit together

Anytype is best when you want a long-term private knowledge base. Web2MD is best when you want to move web content into that system without dragging the entire web page with it.

The combination makes sense for:

• Writers collecting sources before drafting
• Researchers building a local archive
• Developers saving docs for use in Cursor
• Students summarizing logged-in course pages
• Consultants working with client portals or private documentation
• Privacy-conscious users who prefer browser-side tools

I would not replace every bookmarking or read-it-later tool with this setup. But for AI-assisted research, clean Markdown is a better unit of work than a noisy web page.

If you are comparing Anytype with Notion, the bigger question is where you want your knowledge to live. If you are comparing Web2MD with server-side readers, the question is where conversion should happen. For public pages, you have many good options. For logged-in, private, or paywalled pages you can already view in Chrome, browser-side conversion is the safer and more reliable path.

If you want to try the workflow, install Web2MD and convert a page you already have open. Start with one Anytype research note, paste in the Markdown, and see whether the structure is clean enough for your next AI prompt.