Models don't have agency, systems do: how shaped tokens become function calls
sethjuarez · x · 2026-09-29
Seth Juarez's essay "Models Don't Have Agency. Systems Do." dissects where agentic behavior actually comes from.
- Core claim: an LLM only emits tokens — it originates no intent and performs no action, so by the philosophy-of-action definition it has no agency. The runtime built around it does.
- Mechanism: when a model says "Sure, I can help you book a flight to Tokyo Tuesday afternoon," a human can act on it but a program chokes — the words are there, the shape is not. Shaping output as a function call ({ name: "bookflight", ... }) lets the runtime execute the real function; the model still only made text.
- He lays out a progression from word to structure to function: closing the gap by constraining what the model is allowed to say, not by making it smarter — that's what hands agency to the system.
More from coding & agent
- Google Research open-sources PaperOrchestra, a multi-agent framework for writing AI papers — google-research · 2026-11-16
- ffmpeg-skill gives coding agents local video editing via 42 FFmpeg tools and MCP — kajisho5 · 2026-11-16
- Agents Now Reply to Slack Messages on Their Own, Resolving a Prod Incident Ghost-to-Ghost — iamrobotbear · 2026-11-16
- Engineer's Interview Prep Method: Use Claude to Find Which Answers Undersell Your AI Work — brandon_galang · 2026-11-16
- DeepSeek open-sources its app frontend "chat-demo", already at 3.9k stars — NielsRogge · 2026-11-16
- lemire: AI's real impact on coding is expanding the engineer's 'cognitive radius', not speed — lemire · 2026-11-16