ai-weekly-digest

Curated weekly AI landscape digest: top papers, big tech moves, startup funding, open-source releases, and policy developments.

Arguments

This command accepts an optional argument: a target week.

• last — target last week (e.g. /ai-weekly-digest last)
• YYYY-WXX — target a specific ISO week (e.g. /ai-weekly-digest 2026-W08)
• No argument — auto-detect: if today is Monday, target last week; otherwise target the current week.

Examples:

• /ai-weekly-digest — auto (last week on Monday, current week otherwise)
• /ai-weekly-digest last — explicitly target last week
• /ai-weekly-digest 2026-W08 — target a specific week

Step 1 — Determine scope & guard

• Target week: Parse the optional argument:
  • If last → target last week.
  • If YYYY-WXX → target that specific week.
  • If no argument → Monday targets last week, other days target the current week.
• Compute the week identifier as YYYY-WXX (ISO week) and its date range (Monday–Sunday).
• Guard: If 04_Knowledge/AI-Weekly/YYYY-WXX - AI Weekly Digest.md already exists, switch to update mode: read the existing digest and preserve its entries. Steps 2–5 still run to gather fresh data. In Step 6, merge new findings into the existing file — add new items to each section, but do not duplicate entries already present (deduplicate by title/repo name). Inform the user: "AI Weekly Digest for YYYY-WXX already exists — updating with fresh data."

Step 2 — Gather open-source data from vault

Read all 04_Knowledge/GitHub/YYYY-MM-DD - GitHub Trending*.md files whose dates fall within the target week (Monday–Sunday).

• Extract all repos across those files: name, description, stars gained, relevance notes.
• Deduplicate by repo name across days; sum stars gained for repos appearing multiple days.
• Rank by cumulative stars gained.
• Collect all ## Patterns & Themes sections for synthesis in Step 6.

If no GitHub trending files exist for the target week, note this and proceed — the Open-Source section will rely on WebSearch.

Step 2b — Gather academic data from vault

Read all 04_Knowledge/Academic/YYYY-MM-DD - arXiv*.md files whose dates fall within the target week (Monday–Sunday).

• Extract all papers across those files: title, authors, arXiv ID, category, TL;DR, why it matters, relevance notes.
• Deduplicate by arXiv ID or title across days.
• Collect all ## Research Themes & Connections sections for synthesis in Step 6.
• Rank papers by: (1) explicit relevance to vault projects, (2) community buzz mentions from the daily reports, (3) category diversity.

If no arXiv daily files exist for the target week, note this and proceed — the Academic section will fall back to RSS+WebSearch in Step 3.

Step 3 — Fetch RSS feeds

Fetch all of the following feeds in parallel using WebFetch:

#	Feed URL	Section
1	https://blog.openai.com/rss/	Big Tech
2	https://blog.google/technology/ai/rss/	Big Tech
3	https://ai.meta.com/blog/rss/	Big Tech
4	https://www.anthropic.com/rss.xml	Big Tech
5	https://blogs.microsoft.com/ai/feed/	Big Tech
6	https://machinelearning.apple.com/rss.xml	Big Tech
7	https://techcrunch.com/category/artificial-intelligence/feed/	Startups
8	https://the-decoder.com/feed/	Cross-cutting
9	https://www.technologyreview.com/feed/	Policy

arXiv RSS fallback: If Step 2b found no 04_Knowledge/Academic/ files for the target week, also fetch these arXiv feeds to populate the Academic section directly:

• https://arxiv.org/rss/cs.AI

• https://arxiv.org/rss/cs.CL

• https://arxiv.org/rss/cs.LG

• https://arxiv.org/rss/cs.RO

• https://arxiv.org/rss/cs.CV

• Filter entries to the target week's date range (Monday–Sunday).

• If a feed fails, log a warning and continue with the rest.

Step 4 — Filter, rank, and fill gaps with WebSearch

Academic Trends (synthesized from daily reports)

• If Step 2b found arXiv daily files for the target week:
  • Identify 3–5 research trends/themes that appeared across multiple daily reports (e.g. "Efficiency breakthroughs in LLM inference", "New paradigms in robot manipulation").
  • For each trend, note which papers contributed to it and what the collective movement suggests.
  • Select the top 3–5 papers of the week — the most impactful or discussed individual papers across all daily reports.
• If no daily files exist (fallback mode):
  • Use the arXiv RSS feeds fetched in Step 3's fallback.
  • Cross-reference with WebSearch for community buzz (e.g. "best AI papers this week" on Reddit, X, Hacker News).
  • For each selected paper, run one WebSearch for context on its significance.
  • Select 3–5 papers that are most impactful or widely discussed (legacy behavior).

Big Tech (5–8 items)

• Supplement blog feed entries with WebSearch for major announcements not covered by blogs (e.g. "OpenAI announcement this week", "Google AI launch this week").
• Select 5–8 most significant items.

Early-Stage Startups (5–8 items)

• Focus on pre-seed, seed, and Series A rounds — not mega-rounds from established players.
• Supplement TechCrunch and The Decoder entries with WebSearch for early-stage AI startup funding, accelerator cohorts, and new product launches (e.g. "AI startup seed round this week", "AI pre-seed funding this week", "YC AI startups", "AI accelerator cohort 2026").
• Prioritize startups relevant to vault projects (personalization, developer tools, local-first AI, agent infrastructure).
• For each item: company name, stage, raise amount, what they're building, and why it matters competitively or as inspiration.

Landscape & Strategic Context (3–5 items)

• NOT traditional policy/regulation news. Instead, surface items that matter for an early-stage AI founder:
  • Market signals: Shifts in where VCs are deploying capital, emerging consensus on AI product categories, pricing model experiments.
  • Builder-relevant policy: Only regulation that directly affects product design decisions (e.g. data privacy rules that constrain local-first vs. cloud architectures, API ToS changes from foundation model providers).
  • Competitive dynamics: New entrants, pivots, or shutdowns in adjacent spaces (personalization, cognitive AI, agent infrastructure).
• Skip macro governance news (UN summits, EU omnibus amendments, state-level regulation) unless it has direct product implications.

Step 5 — Read active projects for relevance mapping

Read project files in 02_Projects/ with status: active to understand current vault priorities. Use this context to annotate digest items with vault-project connections (e.g. "relevant to [[ProjectName]]", "relates to [[ResearchTopic]]").

Step 6 — Write full digest

Path: 04_Knowledge/AI-Weekly/YYYY-WXX - AI Weekly Digest.md

Create the 04_Knowledge/AI-Weekly/ directory if it doesn't exist.

Frontmatter:

---
type: knowledge
kind: ai-weekly-digest
date: YYYY-MM-DD
week: YYYY-WXX
tags: [ai, weekly-digest, research, industry]
---

Use the actual date (today) and week identifier.

Body structure:

# AI Weekly Digest · YYYY-WXX

## Academic Research Trends

### Weekly Research Landscape
Brief paragraph (3–4 sentences) summarizing the week's academic pulse: how many papers were scanned across the daily reports, which categories saw the most activity, what the overall research direction looks like.

### Key Trends This Week

#### [Trend Name] (e.g. "Efficient Inference at Scale")
**Papers contributing** — [[Paper1]], [[Paper2]], [[Paper3]]
**What's happening** — 2–3 sentences on the research movement.
**Why it matters** — Significance for the field and vault projects.

(Repeat for 3–5 trends)

### Top Papers of the Week

#### [Paper Title](link)
**Authors** — Author list
**TL;DR** — One-sentence summary.
**Why it stood out** — What made this the week's standout (breakthrough result, community buzz, vault relevance).

(3–5 papers, selected from all daily reports)

### Research Radar
- 2–3 bullets on emerging themes that appeared late in the week or in only one daily report but seem significant — "things to watch."

## Big Tech Updates

### Company Name
- **What happened** — Description of the announcement/release.
- **Why it matters** — Impact on the AI landscape.
- **Relevance** — Connection to vault projects, if any.

(Group by company, 5–8 items total)

## Early-Stage AI Startups

### Company Name — $Xm Pre-Seed/Seed/Series A
**What they're building** — One-liner on the product.
**Stage & raise** — Round details.
**Why it matters** — Competitive signal, market validation, or inspiration.
**Relevance** — Connection to vault projects, if any.

(5–8 items, pre-seed / seed / Series A only)

## Open-Source Model & Tool Releases

Top repos this week from [[GitHub Trending]] daily reports, ranked by cumulative stars:

### [owner/repo](link)
**What it is** — One-liner.
**Stars this week** — cumulative gained.
**Relevance** — Connection to vault projects.

(Top 5–8 repos)

### Weekly Open-Source Themes
- Bullet synthesis of patterns across the week's trending repos.

## Landscape & Strategic Context

### Item title
**What happened** — Description.
**Why it matters** — Implication for early-stage AI founders, product design, or competitive positioning.

(3–5 items: VC trends, builder-relevant policy, competitive dynamics — not macro governance)

## Synthesis & Outlook

- **Dominant theme this week** — One paragraph on the week's defining narrative.
- **Biggest signal for vault projects** — What matters most for active projects.
- **Watch next week** — 3–5 items to monitor in the coming week.

Step 7 — Append summary to Weekly Review

Find 01_Execution/YYYY-WXX/Weekly Review.md for the target week.

If it exists, insert a ### AI Weekly Digest block before the ## Reflection heading:

### AI Weekly Digest
- **Research trends:** [top trend] — one-liner
- **Big Tech:** [top signal] — one-liner
- **Startups:** [top signal] — one-liner
- **Open-Source:** [top repo] — one-liner
- **Landscape:** [top signal] — one-liner
→ Full digest: [[YYYY-WXX - AI Weekly Digest]]

If Weekly Review doesn't exist yet, skip this step silently (the digest may be running before /weekly-review).

Step 8 — Confirm

Print:

• The save path of the digest
• The target week (YYYY-WXX)
• One-sentence highlight per section (Research Trends, Big Tech, Startups, Open-Source, Landscape)
• Whether the Weekly Review was updated