HomePlaygroundQel’zyn VIDocsBlogCommunityVideos
--
← Back to blog
Introducing Neural Symbolic Language

Introducing Neural Symbolic Language

NSL TeamFeb 20, 20262 min readAnnouncement

What is NSL?

Neural Symbolic Language (NSL) is a programming language designed from scratch to bridge the gap between symbolic reasoning and neural computation. Where other languages bolt on AI as a library, NSL makes cognitive primitives -- beliefs, observations, causal reasoning, and knowledge graphs -- first-class parts of the language.

NSL uses Python-like indentation syntax with dynamic typing, making it approachable while offering capabilities no other language has.

What Makes NSL Different

  • Cognitive Primitives -- believe, observe, revise, crystallize, and monitor are keywords, not function calls
  • Causal Reasoning -- Express cause-and-effect with causes, because, triggers, and leads_to
  • Knowledge Graph -- Every cognitive operation populates an in-memory knowledge graph queryable with kg.* builtins
  • 40 Standard Library Modules -- From tensor and grad (autograd) to json, http, crypto, tokenizer, py (CPython bridge), memory (74 functions), and more
  • GPU Acceleration -- tensor.matmul() automatically offloads to NVIDIA tensor cores. No configuration needed
  • Pattern Matching -- match/given with literal, range, or-pattern, binding, and guard support
  • Concurrency -- async/await, spawn, channels with send/receive, parallel, and race
  • Two Data Formats -- NSD (text, 3-8x fewer tokens than JSON) and NCF (binary, compressed, HMAC-signed)

Quick Taste

# Variables and types
let name = "NSL"
let pi = 3.14159
let active = true
let items = [1, 2, 3, 4, 5]
let config = {host: "localhost", port: 8080}

# Procedures
:: greet(who) ->
    print(f"Hello, {who}!")

greet("world")

# Lambdas and pipes
let result = items |> filter(fn(x) => x > 2) |> map(fn(x) => x * x)
print(result)  # [9, 16, 25]

# Beliefs with confidence
belief weather = "sunny" confidence 0.8
observe "dark clouds approaching"
revise weather = "cloudy" confidence 0.6

# Causal reasoning
"rain" causes "wet streets"
"wet streets" causes "slow traffic"
let chain = kg.path(kg.find("rain").id, kg.find("slow traffic").id)
print(chain)

Try It Now

Head to the Playground and write your first NSL program. It runs directly in your browser via WebAssembly -- no installation needed. Or chat with Qel'zyn VI, our AI assistant that knows NSL inside and out.

All Posts