Episode 1 · Meet Silan Viking
Silan Hu5 min84% of researchers now use AI to search literature, read papers, draft surveys, and even review. When someone learns about your work through AI, the assistant describes you by what it can read. So research has to be done well, put online promptly and accurately, and you need to know how people and AI actually read it.
Silan Viking is a local workspace built for exactly that. You capture and write on your own machine; it builds the site, publishes it, checks the live version, and shows you how humans and AI crawlers read your work.
What today's options leave out
Common approach | What it is good at | What is missing |
|---|---|---|
Static blogs such as Hexo, Hugo, or Jekyll, including academic themes like al-folio | Free and fully yours | Built for blogging. Papers, projects, and CV are separate pages, every language is written twice, and a template upgrade can break a site you have not touched for months |
WordPress or Notion sites | Quick to start | No notion of how research items relate; SEO plugins only care about Google rankings |
Profiles such as Google Scholar or ORCID | Indexed | A list of papers with no account of what you did; each platform is updated separately |
An analytics script on the page (Google Analytics and similar) | Counts human visits | Most AI crawlers never run scripts, so they barely show up |
What Silan Viking does
- Capture on the spot: tap Catch moment on the desktop app, then type, dictate, or drop in an experiment video.
- Write it up: edit Chinese and English side by side; articles, projects, and CV are linked, so nothing is copied by hand.
- Publish and check: publishing and deploying are two explicit steps, and every deploy is checked against the live site.
- Work with AI: with the skill installed and MCP connected, assistants like Claude can search your content and draft changes, but they can only propose.
- Read the feedback: visits are counted on the server, so you can see which AI crawler read which page and why (training, answering a user, or search indexing). Every article also gets a GEO check that points out what holds AI back.
How it is organised
- Everything you write is Markdown and TOML in a local
content/folder under Git. That folder is the single source of truth. - Pages, sitemap, structured data, and
llms.txtare generated from it and can be rebuilt at any time. - The desktop app, the CLI, and MCP are three ways into the same workspace. A change made in one shows up in the others.
How this series is laid out
- Meet Silan Viking: this episode.
- Install and create a workspace: set up the CLI and desktop app, and keep writing on another computer.
- Capture, then write it up: Catch moment, bilingual writing, and linking.
- Publish, deploy, and check: put content online and confirm the live site matches what you reviewed.
- Let an AI assistant help: the skill, MCP, and reviewing proposals.
- Read the feedback: who reads you, how AI understands you, and what to change next.
Before you start
- The desktop app, Moments, and related features live on the main branch and are installed from source (Episode 2). The public v1.0 release installs with one command but only has the earlier CLI features.
- Local preview needs Docker. Going live needs your own server and a deploy configuration.
- AI features such as dictation, translation sync, and reader review use your own OpenAI or DeepSeek key.
- A crawler request shows that a page was fetched. It does not show that the page was indexed, understood, or cited.
No comments yet