// the find
JetBrains/koog
Koog is a JVM (Java and Kotlin) framework for building predictable, fault-tolerant and enterprise-ready AI agents across all platforms – from backend services to Android and iOS, JVM, and even in-browser environments. Koog is based on our AI products expertise and provides proven solutions for complex LLM and AI problems
Koog is JetBrains' Kotlin framework for building AI agents with tool use, multi-provider LLM support, and graph-based workflows, targeting JVM, JS, WasmJS, Android, and iOS from a single Kotlin Multiplatform codebase. It's for Kotlin/Java teams who want an idiomatic agent framework instead of bolting Python tooling or LangChain4j onto a JVM stack.
Backed by JetBrains with CI running across multiple LLM providers including Ollama, plus a public YouTrack tracker — signals this isn't a side project that goes stale. Mid-conversation LLM switching without losing history is a real, uncommon feature that matters for cost/latency tradeoffs. Native Spring Boot and Ktor integration alongside MCP and ACP protocol support puts it ahead of most JVM agent frameworks on interop. Kotlin Multiplatform targets (iOS, Android, JS/WasmJS) from one codebase is something LangChain4j and Spring AI don't offer.
Ecosystem is a fraction of LangChain4j or Spring AI's size, so expect to write custom tool integrations for anything outside the mainstream providers listed. It's explicitly a JetBrains incubator project and the additions module ships as beta — the 'enterprise-ready' framing in the README is aspirational, not proven by track record yet. Kotlin-first design likely means the Java API is a wrapper rather than a first-class citizen, worth checking before committing a Java-only team. The README's persistence and fault-tolerance claims aren't demonstrated in the quickstart — you have to dig into the docs site to see if retry/state-restore actually hold up under failure.