Content Intake SOP
How Ray (COO) processes articles, posts, videos, newsletters, and other media shared by the founder.
Trigger
Founder shares a link, screenshot, or file via any channel (iMessage, Discord, email).
Process
1. Capture & Read
- Fetch the content (WebFetch for articles, download for attachments, transcribe for video/audio)
- If the source is paywalled or inaccessible, ask the founder to paste the text or share a screenshot
2. Identify the Mental Model
Every good piece introduces a framework or way of thinking. Name it in one sentence. Example: "Hierarchy exists because humans were the only coordination mechanism — AI removes that constraint."
3. Write the Summary
- Summarize in the author's terms first, then in ours
- Weave connections to existing vault content inline using
[[wikilinks]]— don't dump a list of links at the bottom - Connections should read naturally: "This validates the same pattern we're building in [[our skills architecture]]"
- Surface contradictions too — if it challenges something we believe, say so
4. Raise Open Questions
What threads does this material open? What should we be asking that we weren't before?
5. Save to Vault
- Location:
06-reference/YYYY-MM-DD-slug.md - Frontmatter must include:
date— when we received itpublished— when the source was published (captures the zeitgeist)type— article, video, book, newsletter, podcast, tweet, threadsource— URLauthor— who created itshared_via— which channel it came through
6. Update Existing Docs
If the new material meaningfully connects to an active project or SOP, update that doc to reference it. Connections should be bidirectional.
7. Re-index
Run qmd update && qmd embed so the new material is searchable.
8. Confirm Back
Reply in the channel it was shared from with:
- The mental model (one sentence)
- 2-3 most relevant connections to our work
- Any open questions worth discussing
Keep it concise — the full doc is in the vault for deep reading.
Media-Specific Notes
| Media | How to Process |
|---|---|
| Article/blog | WebFetch → summarize |
| Tweet/thread | API fetch or screenshot → summarize |
| YouTube video | Download → transcribe → summarize |
| Podcast | Download audio → transcribe → summarize |
| Book | Founder shares notes/highlights → synthesize |
| Email newsletter | Founder forwards or pastes → summarize |
| RSS feed | Set up /loop to poll → process new items |
| PDF/paper | Read tool → summarize |
Quality Check
A good intake doc should:
- Be readable on its own without clicking any links
- Make non-obvious connections to our existing work
- Name the mental model clearly enough that we can reference it later ("the Block intelligence model", "the capabilities-not-products pattern")
- Include temporal context — what was happening in the world when this was published
- Raise at least one question we hadn't been asking