xqr

xqr aims to make systems and application development dramatically more expressive without giving up access to native performance and existing C/C++ ecosystems. The larger goal is to rethink the programming layer for an AI-assisted world: humans should be able to express architecture and intent at a high level, while agents handle increasingly large parts of implementation, transformation and verification.

Modern software development has accumulated enormous accidental complexity. Developers spend too much time translating intent into boilerplate, navigating framework conventions and maintaining implementation details that add little product value. AI coding agents help, but they still operate primarily on languages and abstractions designed for humans manually writing every line of code. As projects grow, agents also struggle with large context, implicit architecture and the huge space of possible low-level changes.

xqr is an expressive programming language and toolchain that compiles to C++, combining higher-level abstractions with access to native libraries and performance. It is being designed as a better intermediate layer between human intent, AI agents and executable software: concise syntax, explicit semantics, composable abstractions and domain-specific layers can give agents a smaller and more structured space in which to work. The current work includes the language, compiler architecture and runtime/tooling experiments, with the broader goal of building an AI-native development environment around them.