Sign is currently under active development. Specifications and implementation details may change.
NASA Space Apps Challenge Sapporo 2026
Saturday, November 14, 2026 · one of the world's largest global hackathons — in preparation
Sign is a next-generation programming language unifying category theory, zero-cost abstraction, and DPL integration.
It combines keyword-free expressiveness with mathematically grounded design to build a safer software foundation for the age of AI agents.
Sign removes the notion of reserved words entirely. Because syntax is expressed structurally and symbolically, domain-specific vocabulary can be used with far greater freedom.
// Natural expression without reserved words
function calculate_total(items) {
return items.sum()
}
By grounding the language in category theory, Sign aims for stronger type safety and stronger guarantees about program correctness, catching more issues at compile time.
// Categorical composition
f: A → B
g: B → C
h = g ∘ f: A → C
Because domain terms can be used directly, business logic can be expressed more naturally and code becomes easier to read.
// Intuitive business logic
customer.purchase(product)
.apply_discount()
.process_payment()
Sign targets high-level abstraction with zero runtime overhead, aiming for performance on par with or beyond Rust.
// Abstraction with no added cost
items.filter(|x| x > 10)
.map(|x| x * 2)
.sum()
Overview article
Sign Overview is the front door to Sign's ideas, design goals, and structural logic. It is arranged so first-time readers can follow why a new language may be needed, what makes Sign different, and where category theory and the AI era connect to its design.
This is not the raw specification itself. It is the guided entry page that helps first-time readers understand Sign without getting lost in abstraction too early.
Instead of pasting slide fragments, this version is edited into a site-ready overview article for first-time readers.
Development of the Sign language has advanced: the interpreter now generates. The pre-self-hosting stages (preprocess, lexer, parser) are complete, and we have built the local LLM infrastructure.
Next we will train the local LLM and, once training completes, implement the back end of self-hosting. We plan to develop Signify-OS and self-host both the language base and the local LLM, alongside work on a dedicated protocol, applications and UI/UX.
Note: these slides include in-progress results. Performance statements are our own estimates and have not been independently verified.