Groovy as a human and AI friendly programming language

We asked AI agents how to make Groovy 6 more AI friendly, with one rule: every change had to make Groovy more human friendly first. It turns out "AI-friendly" is just "friendly." An agent needs what a new team member needs: to find information without asking anyone, reason about code it hasn't read, and act without breaking things it can't see. This talk covers four layers.

* The language: verified contract annotations such as @Modifies, @Pure, @Ensures and @ThrowsIf turn a method signature into a specification, cutting "what does this touch?" from 460 tokens to 12.
* The toolchain: type-checking extensions catch the mistakes generated code actually makes, including SQL injection, null dereferences, bad regexes and non-associative parallel folds, plus groovy check for a cheap feedback loop.
* The scripting layer: run a script instead of reading the data (305× less context), write MCP servers, and ship them as native binaries that start in 25 ms.
* The project: how apache/groovy works with agents through AGENTS.md, a machine-readable threat model, three safety rules, and treating token budgets as contributor equity.

We also tour Groovy with Ollama4j, LangChain4j, Spring AI, Embabel, Micronaut and Quarkus, and preview Groovy 7 proposals. Every example runs and every number was measured. Presented at Community Over Code 2026, Glasgow.

Transcript