This site is a plain-language guide to understanding AI coding assistant vocabulary. If you’re confused by what “model” means versus “agent” versus “orchestrator” versus “Copilot,” this is for you.

What this site is

This site explains the vocabulary of AI coding assistants. It’s not a product review, not a “how to use Copilot” tutorial, and not a pitch. It’s a mental model and glossary site.

If you’ve found yourself in conversations where:

  • Someone says “AI model” and you’re not sure if they mean GPT-4 or GitHub Copilot
  • “Agent” seems to mean different things depending on who’s talking
  • You hear “MCP tools,” “skills,” “extensions,” and don’t know what binds them together
  • You’re reading docs and the terminology keeps shifting

…then you’re in the right place.

Why it exists

The problem: AI tooling vocabulary is a mess.

The same word—”agent,” “model,” “tool,” “skill”—gets used to mean completely different things by different companies, different products, even different teams within the same company. Developers waste time being confused when they should be building.

This site tries to cut through that by:

  • Showing you the mental model (the AICA Onion)
  • Defining terms clearly in their context
  • Using GitHub Copilot as the primary example (but the patterns apply across the landscape)
  • Linking to official docs so you can go deeper when you need to

How it’s structured

The content is organised around the AICA Onion—a 6-layer mental model for understanding how AI coding assistants work:

  1. Model — the foundation (e.g., GPT-4, Claude)
  2. AI Coding Assistant (AICA) — the product built on top of a model
  3. Agent — autonomous reasoning loops (e.g., Copilot Agent Mode)
  4. Orchestrator — coordination layer (how tasks are routed)
  5. Instructions — your custom rules and context
  6. Tools/MCP — extensions and integrations (what the assistant can do)

Each layer builds on the previous one. Understanding where something sits in the onion helps you understand what it is.

New to the site? Start with Start Here for a quick onboarding, then explore Concepts to build your mental model.

Who made it

This site was built by an engineering director who got tired of being asked questions by people who got confused by AI Coding Assistants over and over. If you find it useful and want to improve it, contributions are welcome. Head to the GitHub repository to open issues, suggest changes, or submit PRs.

A note on accuracy

AI tools move fast. Content here is kept as current as possible, but always verify against official documentation for production decisions. The Docs section links to deeper dives and official sources so you can go directly to the authoritative source when you need to.

What’s next?