Language design
Expressions, results, async, closures, pattern matching and concise domain syntax.
↗ Experimental language · working prototype
A systems language
for building together.
XQR is an expressive programming language and compiler toolchain designed for humans and AI agents to build software together — with a native execution path and access to the C/C++ ecosystem.
01 / Premise
Software has accumulated too much accidental complexity.
Developers translate intent into boilerplate, framework conventions and implementation detail. AI coding agents inherit the same languages and codebases — systems designed for humans manually writing every line.
XQR explores a more structured layer between intent and executable software: concise syntax, explicit semantics, composable abstractions and compiler-based tooling.
02 / Architecture
Intent, made executable.
Not a wrapper around an LLM. A real language architecture designed to make transformation, analysis and verification first-class.
01Source
02AST
03CoreIR
04Passes
05C++ / Native
xqr / pipeline.xqr● prototype
01 flow compile(source) {
02 ast = source.parse()?
03 core = ast.lower(CoreIR)
04 core.passes(verify, optimize)
05 core.emit(Cxx23)
06 }HUMAN + AGENT LOOP
structured
XQR
abstractions
XQR
abstractions
intent
implement
verify
Current path XQR compiles to C++20/23 today. The CoreIR architecture leaves room for an interpreter, VM, JIT and native backends later.
03 / In development
Current focus.
XQR is a working technical prototype under active development — built from language semantics down to runtime experiments.
Compiler architecture
A structured AST → CoreIR pipeline with explicit analysis and transformation passes.
↗Runtime + tooling
Interpreter and runtime experiments, diagnostics, build tools and a path toward native backends.
↗C / C++ interop
Practical access to mature native libraries, performance and the existing systems ecosystem.
↗04 / Direction
One language.
Many systems.
01Expressive systems programming
02Developer tooling, GUI and IDE
03Domain-specific abstractions
04Structured code generation
05Human–agent software development
05 / Builder
EA
Built by a systems thinker.
Evgeniy Arshanskiy is a solo technical founder with 19+ years in software development.
With a background in technical physics and software engineering, he is building XQR across language design, compiler architecture, runtime and tooling — while testing AI agents in real engineering workflows.
06 / Collaborate
Let’s build
the layer between.
Looking for early collaborators, technical discussions and a product-minded co-founder interested in compilers, developer tools and human–agent software development.
Start a conversation ↗evgeniy.arshanskiy@gmail.com