What is mattwood.fyi?

A live, structured feed of riffs and links from Matt Wood, Chief AI & Technology Officer at AWS — readable by people and queryable by agents as a typed knowledge graph.

Is there an API and is it free?

Yes. A public, read-only query API is available with no authentication and no API keys. See https://mattwood.fyi/openapi.json and https://mattwood.fyi/docs.

How should an AI agent use this site?

Fetch https://mattwood.fyi/feed.json for the full structured feed, or query the API for keyword/semantic search and the connection graph. See https://mattwood.fyi/agents/ for when-to-use guidance.

WorldTUI: Training a 360M Model as a Terminal Emulator

A developer fine-tuned a 360M-parameter language model to simulate a terminal by predicting screen changes from keystrokes, without an actual shell or filesystem behind it. The model can run vim, tmux, and other programs, with the ability to self-correct and maintain consistent state across commands.

I don't even know how to think about this.

simedw.com →

Connections

Related to

ProcInSh: Web-based Linux Process Inspector ProcInSh is a web-based Linux process inspector and WorldTUI simulates terminal behavior - both reimagine how terminal/shell environments can be abstracted and presented through non-traditional means
OSC 7501: A Terminal Protocol for Program Status OSC 7501 is a terminal protocol for program status reporting; WorldTUI simulates terminal screen state prediction - both work at the level of terminal state representation and communication
Language Models that Play Chess and Explain Their Moves Both demonstrate LLMs learning domain-specific interactive behavior (chess moves vs terminal commands) through prediction rather than rule execution, showing models can internalize procedural interaction patterns
Tuios: Terminal Window Manager for AI Agents Tuios is a terminal window manager designed for AI agents, while WorldTUI simulates a terminal via AI - both explore the intersection of terminals and AI, from opposite directions

Supports

Dust: Pretraining Transformers Without Backpropagation Dust trains transformers without backpropagation; WorldTUI's 360M model achieving complex terminal simulation supports exploration of unconventional training approaches for behavioral prediction tasks

Challenges

The Slow Formation of Durable Software The Slow Formation of Durable Software argues software needs stable foundations; WorldTUI challenges this by replacing a deterministic shell backend with probabilistic model predictions, questioning what 'running' software even means

Dessa: Conversation Model Trained on Speech Both explore unconventional training data sources for language models — WorldTUI uses terminal emulator interactions while Dessa uses speech audio, both pushing beyond standard text corpora

Type:	link
First published:	Thursday, 22 October 2026 at 16:12 UTC
Location:	https://simedw.com/2026/10/21/world-model-terminal/
URL:	https://mattwood.fyi/i/0102e869

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.

Connections