Kotlin
A concise multiplatform language developed by JetBrains
KotlinConf’26 Keynote Highlights: Advances in Language Design, Tooling, AI-Driven Workflows, and Multiplatform Development
Kotlin turns 15 this year, and it really is everywhere. It powers systems behind everyday moments, such as tapping to pay, buying commuter rail tickets, using in-flight entertainment, and even filing tax returns online. As AI continues to reshape how software gets built, Kotlin’s growing real-world impact reflects the importance of languages and tools that help teams manage complexity, express ideas clearly, and build reliable systems with confidence.
At KotlinConf’26, the JetBrains team and industry partners shared how Kotlin continues to evolve for developers at every scale. The keynote highlighted advances in language design, tooling, AI-driven workflows, and multiplatform development – all aimed at improving the Kotlin development experience for building modern applications everywhere.
Evolving Kotlin
As AI-driven development raises the level of abstraction, trust in the programming language is more important than ever. Kotlin Lead Language Designer Michail Zarečenskij mentioned that with Kotlin, the team aims to provide that trust at every level. Ergonomics and safety are principles that guide the language at its very core.
Michail previewed Kotlin 2.4.0 – the next step in Kotlin’s evolution toward safer and more ergonomic code. Among the features being stabilized are context parameters, designed to make APIs more expressive and focused on core logic, and explicit backing fields, which simplify common backing property patterns while reducing boilerplate and improving safety.
The presentation also covered several experimental language features, including multi-field value classes for modeling domain-specific data such as money or colors. Key aspects of value classes include:
- The compiler automatically generates functions like
equals(),hashCode(), andtoString(). - Value classes use safer name-based destructuring by default.
- Value classes have no identity semantics – they are fully defined by their properties.
These changes are designed to make working with data safer, more expressive, and more efficient over time.
Among other experimental features, the presentation highlighted updates to collection literals, locality as a first-class language concept, and rich errors, a new approach to representing and handling recoverable failures.
Kotlin ecosystem
Tooling has been part of Kotlin’s story from the beginning. As Kotlin expands into new workflows, including agents and integrations, the ecosystem continues to apply the same core principles of ergonomics and safety. The goal is to ensure a consistent development experience with any editor, build tool, or agentic framework.
One of the major announcements was the Kotlin Toolchain – a unified entry point into the Kotlin ecosystem. Available through a single command, Kotlin Toolchain brings together everything from creating, building, running, and testing applications to formatting code, generating documentation, and integrating with agents.
Starting today, you can already use Kootlin Toolchain in your JVM and multiplatform projects to build, run, and test your apps, with Amper now serving as the core part of the Kotlin Toolchain. In the future, Kotlin Toolchain will expand with LSP integrations, AI skills, native dependency provisioning, and much more. As always, JetBrains is also bringing deep IDE integrations for the best possible out-of-the-box experience.
The presentation also introduced the Kotlin Documentation Model, a core part of Kotlin that represents machine-readable documentation in the form of a kdoc.jar. This specified, backward-compatible format will be published alongside libraries and consumed by IDEs, web tools like Dokka, and AI agents.
Another major announcement was the promotion of the Kotlin Language Server to Alpha. Backed by the full power of the IntelliJ engine, LSP provides a more consistent experience across diagnostics, code completion, and tooling support. The official Kotlin extension for Visual Studio Code is now also available on the Visual Studio Marketplace.
As part of Kotlin Foundation’s efforts, JetBrains and Meta have started the process of standardizing ktfmt and making it a core part of Kotlin.
The team also announced ongoing collaboration with the open-source community to bring first-class Kotlin support to official Bazel rules_kotlin, making it easier to use Kotlin in large-scale codebases with thousands of modules.
Kotlin at Google
Google has been using Kotlin in production for over a decade, and 92% of professional Android developers now use Kotlin for Android applications.
The keynote also highlighted Google’s ongoing collaboration with JetBrains on the K2 compiler. Since launching stable K2 support in Android Studio, the Google team has seen near-universal adoption. In Kotlin Symbol Processing, Kotlin’s solution to Java annotation processes, which Google built and maintains, a 17% reduction in execution time for complex builds was achieved. In R8, Android’s whole-program optimization tool, the team rewrote coroutine locks to avoid reflection, saving up to 50% on composed performance benchmarks.
AI tooling for Kotlin
The keynote also focused on the next generation of AI tools for Kotlin development. We want you to be able to use any agent directly inside JetBrains IDEs. To support these efforts, JetBrains is co-leading the development of an open standard, the Agent Client Protocol (ACP), which specifies how IDEs and coding agents communicate. You can read more about it in our dedicated blog post: Our 2026 Direction: AI and Classic Workflows in JetBrains IDEs.
Junie
Junie, JetBrains’ coding agent, is deeply integrated with JetBrains IDEs, and even the Junie CLI version can connect to the IDE to get full project context. Junie also works with different LLM providers, allowing you to choose the best model for a specific task. While Junie already works in Kotlin projects, it now also includes dedicated Android support.
JetBrains Air
As developers are becoming more productive with agents, the keynote also explored how to scale agent-based development workflows. JetBrains Air is an agentic development environment for working effectively with multiple agents.
OpenAI Codex, Claude Agent, Gemini CLI, and Junie can execute independent task loops without conflicting with one another. You can start agents in separate Git worktrees or Docker containers, and to share progress with your entire team, you’ll soon be able to use cloud agents and even start and guide them directly from the browser.
Koog 1.0
Vadim Briliantov, Technical Lead and author of Koog, continued the keynote with a talk about the Kotlin AI agent framework that allows you to build fault-tolerant, scalable, and enterprise-ready AI agents in fully idiomatic Kotlin. Vadim announced the stable release of Koog 1.0, a major milestone for production-ready agent development in Kotlin across backend, mobile, and multiplatform applications.
The presentation focused on Koog’s approach to building reliable AI systems through type-safe workflow DSLs, persistence and recovery for long-running agents, and deep integrations with existing Kotlin ecosystems, including Spring AI, Ktor, and observability tooling. One of the featured case studies came from Mercedes-Benz, whose team uses Koog to build vehicle maintenance support agents with structured workflows and carefully controlled execution logic. Read the full case study.
Vadim also showcased multiplatform support and on-device AI capabilities on Android using Google’s Gemma models, reinforcing Kotlin’s growing position as a unified language for building modern AI-powered applications – from backend services to mobile experiences – all in Kotlin.
Kotlin for backend development
The keynote continued with updates on Kotlin for backend development, including new capabilities across Ktor, kotlinx-rpc, and Exposed. The team showcased Koog integration for building AI-powered services with Ktor, experimental first-party gRPC support in kotlinx-rpc, and the stable release of Exposed, which introduces vector types for AI-powered similarity search alongside a new Gradle plugin for simplified migration script generation. A new agent skill is also available to help developers migrate existing projects to Exposed 1.0.
Beyond tooling, the presentation focused on Kotlin’s growing adoption in enterprise and compliance-driven environments, where reliability and long-term support are critical.
Starting with Kotlin 2.4, the Kotlin standard library will include an 18-month security support policy, with security fixes backported to all release lines within an active support window.
The keynote also highlighted Kotlin’s productivity benefits for backend teams, referencing data showing 15–20% faster development cycles as projects grow in complexity.
The presentation emphasized Kotlin’s deep integration with the JVM ecosystem through ongoing collaboration with Spring, improved Kotlin representation in Spring and JUnit documentation, updates to the kotlin-maven-plugin and Maven onboarding experience, improved coroutine support in Micrometer, and continued stabilization of the Lombok compiler plugin for mixed Kotlin-Java projects.
Kotlin Multiplatform
Kotlin Multiplatform continues to see rapid adoption, with the number of top apps using KMP more than doubling over the past year. Companies such as PayPal, Booking.com, Sony, and Duolingo are already using it in production, and more teams are adopting Compose Multiplatform to share UI across platforms.
For example, Sony uses KMP in their Sound Connect app for headphones to work with platform APIs like sensors and background processing while sharing the UI through Compose Multiplatform. Across Kotlin Multiplatform case studies overall, applications built with KMP now serve hundreds of millions of users daily.
Getting started with KMP is now easier with the KMP IDE plugin available on all operating systems for both IntelliJ IDEA and Android Studio. The plugin offers everything you need to build great KMP apps with convenient run configurations, tools for working with Compose code, integrations with Swift and cross-language features, and AGP 9.0 support.
You can also create new projects right in the IDE with the KMP project wizard, which now uses our new default structure, where each module has a single clear responsibility.
We are working to enhance the iOS development experience, notably through Swift Export features that make calling Kotlin from Swift more natural. In Kotlin 2.4, Swift Export is officially moving to Alpha. We also introduced SPM import, which lets you add dependencies on Objective-C-compatible code using Swift Package Manager and call into those APIs directly from Kotlin code.
Kotlin/Native has seen significant performance improvements over the last year. Measured on the Google Docs codebase, build times are now 25% faster while using less than half the RAM during builds compared to a year ago.
Compose Multiplatform
Compose Multiplatform is fully stable and production-ready on mobile and desktop. The web platform also reached Beta status in September 2025, marking another major step forward for multiplatform UI development. For all these platforms, the team continues to bring you the latest improvements and APIs from Jetpack Compose. One of the biggest highlights over the past year is the new Navigation 3 library, a flexible, Compose-first solution that gives you full control over your back stack – and it’s already stable for multiplatform use.
On iOS, new interop APIs now make it possible to combine native Liquid Glass components with Compose UI, allowing native views to dynamically interact with Compose content underneath.
Beyond the framework itself, the Kotlin Multiplatform ecosystem continues to expand rapidly. There are now more than 3,500 community libraries listed on klibs.io, giving you a growing set of tools and integrations for building multiplatform applications across mobile, desktop, backend, and web.
Conclusion
KotlinConf’26 highlighted how Kotlin continues to evolve beyond a programming language into a complete ecosystem for backend, mobile, web, AI, and multiplatform development. From language and tooling improvements to growing industry adoption, the announcements reflected the shared goal of helping developers build modern software with greater clarity, safety, and productivity.
