v0.1 · Pre-alpha — the compiler is under active development Source on GitHub ↗
Design goals

Twelve decisions behind the language

Each of these is a deliberate trade-off, not a checklist of buzzwords. Items marked Future are on the roadmap but not shipped.

01

Multilingual programming

Native keywords in Tamil, Tanglish, English, with more Indian languages on the roadmap.

02

AI-friendly syntax

Predictable, unambiguous grammar designed for accurate generation and static analysis by AI models.

03

Native compilation

Compiles through LLVM straight to machine code — no interpreter, no VM overhead.

04

Strong static typing

Types are checked at compile time with clear, actionable diagnostics.

05

Cross platform

One toolchain targeting Windows, Linux, macOS, and beyond.

06

LLVM powered

Built on a battle-tested optimizing backend used by Rust, Swift, and Clang.

07

Memory safe

Immutable-by-default design with a borrow checker planned for v1.0.

08

Beginner friendly

Learn programming concepts in the language you already think in.

09

Enterprise ready

Designed from day one for predictable performance and long-term maintainability.

10

Voice-friendly programming

Keywords are chosen to be easy to speak and dictate aloud, for coding by voice in Tamil, Tanglish, or English. Future

11

Natural error messages

Compiler diagnostics read like plain Tamil (or your chosen language), not cryptic English stack traces.

12

AI language translation

Convert TEN source between language packs — write in Tamil, generate the Hindi or English version. Future

TEN vs the field

How TEN compares

FeatureTENPythonRustGo
Multilingual syntaxYesNoNoNo
AI-native grammarYesPartialNoNo
Native compilationYesNoYesYes
Static typingYesNoYesYes
Memory safetyYesPartialYesPartial
Cross platformYesYesYesYes
Under the hood

Compiler architecture

Select any stage to see what it does. Source flows top to bottom — language-agnostic from the lexer onward.

Language packs

Built for India, designed for the world

Built for the AI era

Programming for the AI era

TEN's grammar is unambiguous by construction, so AI systems can read, generate, and optimize TEN code with far higher accuracy than languages designed for human convenience alone.

AI-readable syntax

Clear token boundaries and minimal ambiguity make parsing reliable for models.

Predictable grammar

No hidden context-sensitivity — what you see is what gets parsed.

AST-aware compiler

Tooling can inspect and transform the AST directly, built for codegen pipelines.

Semantic-friendly language

Type information stays close to syntax, easing static reasoning.

Local AI integration

On-device code assistance is on the long-term roadmap. Future

Canonical AST

Every dialect — Tamil, Tanglish, English — compiles to one shared AST.

Cross-language translation

The canonical AST lets AI models re-render any TEN file into another language pack with no logic drift. Future

Localized diagnostics

Errors, warnings, and hints are generated in the dialect you're coding in, not translated after the fact.

Reliability by default

Safety

Immutable by default

Variables don't change unless explicitly marked mutable.

Type checking

Errors caught before your program ever runs.

Compile-time diagnostics

Clear, specific error messages — not stack traces.

Borrow checker Planned

Rust-inspired ownership model targeted for v1.0.

Safe concurrency Planned

Data-race-free concurrency primitives on the roadmap.

Predictable errors

No silent type coercion, no surprising runtime behavior.

Honest numbers

Performance

TEN is pre-alpha. Real benchmarks will be published once they exist — no projected or simulated numbers.

Compilation speed In Progress

Benchmarked against Python, Java, Go, and Rust once the LLVM backend lands.

Execution speed In Progress

Awaiting compiler completion before measurement.

Memory usage In Progress

Will be tracked across release binaries.

Who's building this

Team

AY

Anbuselva Y

Founder, Owner & Compiler Lead
Where TEN is headed

Roadmap

v0.1 — Foundation
  • Lexer
  • Parser
  • Semantic analysis
v0.2 — Type System
  • Type checker
  • Error recovery
v0.3 — Native Codegen
  • LLVM backend
  • Optimization passes
v0.4 — Reach
  • Cross-platform binaries
  • Package manager
v1.0 — Stable Release
  • Borrow checker
  • Safe concurrency
  • Public 1.0 binaries
v2.0 — AI-Native Compiler
  • Local AI code assistance
  • Model-optimized tooling