Agents are monads (but not that kind)
The author argues that AI agents can be understood through the programming concept of monads, specifically the "do-notation" style of composition. They explore how monads provide a structured way to model agent behavior and state management, drawing parallels between functional programming patterns and agent architectures.
Background
- Xe Iaso is a well-known software engineer and blogger who writes about Rust, programming language theory, and systems design. This post appeared on their personal blog.
- A **monad** (in programming) is a design pattern that wraps a value in a context and provides two operations — one to put a value into that context, and one to chain operations while preserving the context. Monads are central to Haskell and functional programming, and have a reputation for being confusing to learn.
- There is an ongoing — often semi-joking — online tradition where programmers write "monad tutorials" that inevitably explain monads by referencing **endofunctors** (a category theory concept). The opening line of this post plays on that trope.
- The broader context: people have been debating whether LLM-based "agents" (autonomous AI systems that decide their own next steps) are a genuinely new paradigm or just a rebranding of old concepts. By comparing agents to monads, the post makes a technical argument about how agent systems compose and sequence operations.