| JetBrains Makes Kotlin LLM Open Source |
| Written by Mike James | ||| | Monday, 03 August 2026 | ||| | JetBrains has announced that KotlinLLM, its research prototype for delegating runtime logic to a large language model (LLM) from Kotlin code, will now be available as open source and will be public. Kotlin is the open source language originated by JetBrains and promoted by Google as a first class language for Android to replace or supplement Java. Kotlin LLM, which was announced in July, is an IntelliJ IDEA plugin prototype. It adds a language feature JetBrains calls smart macros, which can be used for code generation, runtime updates, and hot-re. The GitHub page for KotlinLLM says a smart macro is an explicit Kotlin call whose behavior is backed by generated Kotlin source code. When the running application reaches an unsupported scenario, the plugin can capture the runtime values, ask an LLM agent to generate a narrow implementation update, compile the changed source, and hot-reload the affected class through JDI. The developers say that the design goal is that LLM-backed behavior should be explicit at the call site, persistent as source code, and portable as ordinary Kotlin once generated. LLMs are used more generally during code development for code completion, code generation, and program comprehension. Using an LLM at the runtime of a compiled application is less common. KotlinLLM is built to be explicit, persistent, and portable. The explicit aspect means the call site shows that a feature is LLM-backed, so it's visible in code review. It is persistent because the generated behavior is saved as an ordinary Kotlin source, not kept only in the runtime session. It can be committed, reviewed, tested, and distributed like any other code. And it is portable because once generated, the code runs as plain Kotlin without the plugin. KotlinLLM has three layers, a public API, a static infrastructure and a dynamic infrastructure. The public API consists of two functions, one to convert a value from one type to another, and a second that creates an implementation of an interface whose behavior can evolve from runtime interactions. The static infrastructure consists of managers that route smart macro calls to generated providers, while the dynamic infrastructure consists of a generated bootstrap, provider, parser, and mock classes maintained by the plugin. Unlike direct runtime delegation to an LLM service, KotlinLLM does not call the model on every smart macro invocation. The LLM is used when generated code cannot handle a scenario. Successful behavior is stored in generated Kotlin files and later runs execute that code directly. KotlinLLM is available now. The repository contains the KotlinLLM IntelliJ IDEA plugin, the smart macro API, and example projects.
More Information## Related Articles #
To be informed about new articles on I Programmer, sign up for our
Comments #
or email your comment to: | ||| | Last Updated ( Monday, 03 August 2026 ) |