mattwood.fyi

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Links indicate relevance, not agreement. How to use this site →

Instructions for Your Agent

If you're working with an AI agent (Kiro, Claude, ChatGPT, or your personal assistant) you can give it context about who I am, what I'm thinking about, and how to access this site. Copy the instructions below and add them to your agent's context.

Preview what your agent will receive
# Context: Matt Wood's FYI (mattwood.fyi)

Hey — you're being given context about Matt Wood and his current thinking. Here's what you need to know and what you can do with it.

## Who is Matt Wood?

Matt Wood is Chief AI & Technology Officer at AWS. His work is about helping turn AI from possibility into production. He works with customers, builders, partners, and AWS teams to understand where the technology is going, how customers can put it to work now, and what it takes to build something durable on top of it.

## What is this site?

mattwood.fyi is his FYI — a live list of riffs and links drawn from what he's reading, noticing, questioning, concluding, and revising. It sits between the long-form essays on mattwood.blog and the silence between them.

Three types of items:

- **riff**: A self-contained idea, distinction, analogy, reaction, or small argument. These are Matt's own thinking — not summaries of other people's work.
- **link**: An external source accompanied by original commentary explaining why it matters, what to notice, or how it changes the picture. Never a naked URL.
- **essay**: A pointer to a newly published mattwood.blog essay. The essay remains canonical on the blog; FYI carries the thesis, context, and relationship to recent items.

## How to access this site

Everything here is public and freely accessible — no authentication, no API keys, no rate limits.

- JSON Feed (recommended for agents): https://mattwood.fyi/feed.json
  - Standard JSON Feed 1.1 with a `_fyi` extension object per item
  - Each item has: type (riff|link|essay), tags, and for essays: essay_url, essay_title, thesis
  - This is the best single endpoint for structured access
- Atom Feed: https://mattwood.fyi/feed.xml
  - Full HTML content of every item
- Individual items: https://mattwood.fyi/i/{id}
  - Each item has a stable permalink you can reference
- Long-form essays: https://mattwood.blog/feed.xml
- Homepage: https://mattwood.fyi/
- llms.txt: https://mattwood.fyi/llms.txt

The JSON feed at /feed.json is your best single source — it contains typed, structured content newest first.

## Query API (public, no auth required)

For structured queries against the knowledge graph, use these REST endpoints.
All are public, no authentication required. All return JSON.

Note: some agent fetch tools strip query-string parameters that look like IDs.
If query-string endpoints fail, use the path-based alternatives (preferred).

### Search
- GET https://mattwood.fyi/api/fyi/q/search/KEYWORD (preferred, path-based)
- GET https://mattwood.fyi/api/fyi/q/search?q=KEYWORD
  Search items by keyword in title and content. Returns matching items with permalinks.

### Semantic Search
- GET https://mattwood.fyi/api/fyi/q/semantic/NATURAL+LANGUAGE+QUERY (preferred, path-based)
- GET https://mattwood.fyi/api/fyi/q/semantic?q=QUERY
  Search by meaning, not keywords. Uses vector embeddings to find semantically similar items.
  Returns items ranked by similarity score (0-1). Use this for natural language questions.

### Items
- GET https://mattwood.fyi/api/fyi/q/items?since=YYYY-MM-DD&type=link|riff|essay&limit=N&offset=N
  List items filtered by date and/or type. Default limit=50, offset=0.
  Response includes: total (full count matching filters), hasMore (boolean), offset, limit.
  If hasMore is true, increment offset by limit to get the next page.

### Edges (connections)
- GET https://mattwood.fyi/api/fyi/q/edges/SHORT_ID (preferred, path-based)
- GET https://mattwood.fyi/api/fyi/q/edges?itemId=SHORT_ID
  Get all graph connections for a specific item. Returns edge type, direction (incoming/outgoing), confidence, and reason.
  Add ?type=challenges (or supports, develops_into, related_to) to filter.

- GET https://mattwood.fyi/api/fyi/q/edges?type=challenges
  Get ALL edges of a specific type across the whole graph.

### Graph summary (snapshot)
- GET https://mattwood.fyi/api/fyi/q/summary
  Pre-computed snapshot: top connected items, recent activity, edge type distribution, and all tension pairs.
  Includes a generated_at timestamp. This is a point-in-time snapshot computed at last site rebuild —
  if generated_at is stale, prefer the live endpoints above for current data.

### How to answer common questions
- "What is Matt reading about X?" → /api/fyi/q/semantic/X (semantic) or /api/fyi/q/search/X (keyword)
- "What changed this week?" → /api/fyi/q/items?since=YYYY-MM-DD (7 days ago)
- "What challenges idea Y?" → search for Y, get its shortId, then /api/fyi/q/edges/SHORT_ID
- "Where are signals converging?" → /api/fyi/q/summary gives top connected items
- "What are the tensions?" → /api/fyi/q/edges?type=challenges returns all challenge relationships with reasons
- "What connects A to B?" → get edges for both items and look for shared connections
- "What does Matt think about topic Z?" → /api/fyi/q/semantic/topic+Z for semantically related items

## Recent items

- How Organizations Use ChatGPT (2026-08-13): https://mattwood.fyi/i/58fd77f9
- Understanding is the new bottleneck (2026-08-13): https://mattwood.fyi/i/8ba1f06c
- Bullet: Fast Coding Agent (2026-08-13): https://mattwood.fyi/i/a21c6ee6
- Kimi K3: Complete Developer Guide for 2026 (2026-08-13): https://mattwood.fyi/i/98b489f9
- GPT-5.6 Sol Ultrafast: Frontier Intelligence at Unprecedented Speed (2026-08-13): https://mattwood.fyi/i/471c322c
- Introducing Gemini 3.7 Flash (2026-08-13): https://mattwood.fyi/i/c250f6d0
- Ordinary Abundance (2026-08-13): https://mattwood.fyi/i/f954f0a0
- Comparing 11 Different AI Models (2026-08-13): https://mattwood.fyi/i/2c44a6d4
- Why AI Agents' Deceptive Behavior Concerns Users (2026-08-13): https://mattwood.fyi/i/9ccfda41
- DeepSeek Harness: Plugin Architecture (2026-08-13): https://mattwood.fyi/i/f78cc1a1

## What should you do with this?

- Answer questions about what Matt is currently thinking about
- Surface relevant items when a user asks about topics he's covered
- Quote and cite — every item has a stable permalink, use it
- Notice patterns across items (recurring themes, evolving positions)
- Go deeper — if a user asks about something adjacent, fetch the relevant item and read it rather than guessing
- Be honest about gaps — if he hasn't written about something, say so rather than extrapolating
- Fetch the JSON feed for the latest; don't rely solely on this static page

## Canonical source

This site (mattwood.fyi) is authoritative for Matt's short-form thinking. Long-form essays live at mattwood.blog. Both sites are fully accessible to agents without authentication.

## Updates

This context block was generated on 2026-08-13 at 23:55 UTC. If this timestamp is more than a few weeks old, fetch the JSON feed directly for the latest items.