▶ AI-Native · Non-Commercial Open Source · Eulyph LLC

Eulyph

Bridging Human Intent and Machine Execution
A brand-new programming language where AI isn't an add-on — it's baked into the syntax. Write what you mean, not how a computer needs to hear it. Eulyph handles the thinking so your team handles the results.
🔒 Non-Commercial Free License ✨ Commercial License Available 🐍 Python-powered Runtime
example.eulyph
# Load a local AI model
ai = model("http://localhost:1234/v1")
ctx = ai.context()
 
# @ asks the AI — no API boilerplate needed
review = ctx @ "Summarize this report"
 
# ? -> routes by meaning, not exact keywords
(review ? "Action required?" -> urgent_queue
             else: archive)
 
# Serve any result as a live web API
ctx.serve()
> API live at http://127.0.0.1:5000  

// what is eulyph
A language built for the AI era
Traditional programming languages require exact instructions — one wrong keyword and the whole program breaks. Eulyph is different. It understands meaning, not just characters, making automation that was previously impossible suddenly straightforward.
Fluid track — AI-driven

Natural language operations

Ambiguous decisions, data extraction, classification, and reasoning — handled natively by the @ and ? -> operators via a local LLM. No API wrappers, no boilerplate.

@ inference ? -> branching ctx memory LLM-powered
Rigid track — Precise execution

System-level logic

Math, functions, file I/O, Python module imports, JSON handling, and live web API serving — all handled through the secure Standard Library with 100% precision.

std library extern Python native JSON .serve() API
🧠

AI at the core

The operators are first-class syntax — not library calls. The language itself understands intent.

Radically less code

Replace hundreds of rigid if/else rules with a handful of intent-driven lines. Complexity condenses into clarity.

🛡️

Secure by design

Raw execution is hidden behind safe wrappers. Every AI call is sandboxed. Your data stays local.

🌐

Instant web APIs

Any object or context becomes a live Flask endpoint with .serve() — zero configuration.


// core operators
Two symbols that change everything
Instead of verbose logic, Eulyph gives you two first-class operators that speak directly to the AI engine — replacing thousands of lines of brittle code with a single line of intent.
@

The Inference Operator

Ask the AI anything — get a direct answer

Hands any data to the AI and returns a result. No API wrappers, no JSON parsing, no HTTP boilerplate. Just describe what you want in plain language and Eulyph handles the rest.

# Extract price from a messy email
price = raw_email @ "total cost in dollars"
> $2,500.00
 
# Summarize a document
summary = report @ "key findings"
? ->

The Branching Operator

Route by meaning, not exact keywords

Replaces entire chains of if/else logic with a single AI-powered decision. Judges by intent — "smashed screen" routes correctly just like "broken screen" — without any hard-coded rules.

# Classify and route in one line
(ticket ? "Is this urgent?" -> priority_queue
             else: standard_queue)
> priority_queue
// additional language features

function + return

Full user-defined functions with parameters, closures, and return values — reusable logic anywhere in your program.

extern keyword

Import any Python module directly: extern json, extern requests. Full access to Python's ecosystem from inside Eulyph.

: clone operator

Creates a deep copy of any object — contexts, environments, JSON — so you can branch logic safely without mutating the original.

env() sandbox

Spin up a Python execution environment from within Eulyph. py.exec() and py.eval() run code in an isolated, secure sandbox.


// the codebase — built from scratch
Eight files. One complete language.
Eulyph is hand-crafted with no language framework shortcuts. Every component — from tokenizer to runtime — is purpose-built to handle both the AI "fluid" track and the Python "rigid" track seamlessly.
lexer.py
The Reader

Tokenizer

Scans raw .eulyph source using regex patterns and breaks it into a clean token stream — words, strings, symbols like @, ?, ->, and :. Whitespace and comments are stripped.

parser.py
The Organizer

Recursive Descent Parser

Takes tokens and builds an Abstract Syntax Tree using recursive descent — handling everything from simple variable assignments to complex chained AI expressions and function definitions.

ast_nodes.py
The Blueprints

AST Node Definitions

Defines every instruction Eulyph understands: Assignment, QuestionExpr, FunctionDefinition, CloneNode, ExternNode, and more — the exact structural forms the parser fills out.

evaluator.py
The Executor

Tree-walking Evaluator

Walks the AST node by node. Manages variable scope and closures, routes @ calls to the local LLM, handles ? -> branching decisions, and executes built-in functions like readfile() and writefile().

runtime.py
The Workspace

Runtime & Environment

Provides the Environment class (Python sandbox), the Model and Context objects, and the serve_object() function that spins up any object as a live Flask web API with .serve().

types.py
The Filing System

Native Type System

Defines Eulyph-native types: Text (strings), URL (auto-validated — strings starting with https:// auto-promote), and JSON (native dict/list with bracket access).

client.py
The Phone Line

AI Network Client

The direct networking link between Eulyph and the local LLM endpoint. Every @ operator routes here — building the prompt, calling the model via HTTP, and returning the response to your program.

main.py / __init__.py
The Front Door

Entry Point

Ties the package together. Run any .eulyph file with a single terminal command — it lexes, parses, and evaluates end-to-end. __init__.py makes the whole folder a proper Python package.


// case study — active deployment
Eulyph in action: IT Helpdesk Automation
📌 Custom Script · Not Part of Core Download

The following is a real Eulyph script built specifically for an IT department demonstration — showing how Eulyph can be adapted to any team's needs. This script is not part of the base Eulyph language; it's an example of what you can build with it.

💡

The Problem

Traditional IT scripts rely on exact keyword matches. A ticket saying "smashed screen" instead of "broken screen" breaks keyword-based routing rules entirely — requiring constant manual updates.

🤖

The Solution

A Eulyph script uses @ to read each ticket and ? -> to route it by meaning — not keywords. "Smashed screen" routes correctly just like "broken screen" does. Every time.

The Impact

Eulyph acts as an always-on routing engine. It classifies tickets, suggests troubleshooting actions, extracts search terms, and exports results — replacing hours of manual sorting.

ctx

AI Memory / Context Object — Used in this IT Script

In this helpdesk script, ctx is an AI context object that wraps the local model and holds conversation history. It's initialized once with the department routing library, then reused for every ticket — so the AI already knows the rules without being re-briefed for each request.

# Connect to local model and create a persistent context
routerModel = model("http://localhost:1234/v1")
ctx = routerModel.context()
 
# ctx remembers the dept_lib across all tickets — no repetition
analysis = ctx @ "Review this IT ticket: '" + description + "'..."
// live_email_dispatch.eulyph — actual source
live_email_dispatch.eulyph
# Set up Python sandbox for JSON + dept library
py = env()
py.exec("import json")
dept_lib_str = py.eval("str(dept_lib)")
 
# Connect to local AI model
routerModel = model("http://localhost:1234/v1")
ctx = routerModel.context()
 
# Define the ticket processor function
function process_ticket(description, ticket_id, expected_dept) {
  analysis = ctx @ ("Review: '" + description + "'. Compare against dept library...")
 
  assigned_dept = (analysis ?
    "Confidence > 80?" -> ctx @ ("Extract dept from: " + analysis)
    else: "IT helpdesk team")
 
  confidence = ctx @ ("Extract confidence score from: " + analysis)
  reason = ctx @ ("Why was this routed to " + assigned_dept + "?")
  return "Success"
}
 
# Export results to JSON report
py.exec("json.dump(test_report, f, indent=4)")
// terminal output
$ python main.py
► Processing ticket #1...
  Dept:     Break-Fix team
  Action:  Check display cable
  Terms:   screen, display, broken
  Confidence: 94
 
► Processing ticket #2...
  Dept:     Infrastructure team
  Action:  Restart network adapter
  Terms:   wifi, vpn, disconnected
  Confidence: 88
 
► Processing ticket #3...
  Dept:     Apps team
  Action:  Reinstall application
  Terms:   outlook, crashing, 365
  Confidence: 91
8 Departments · 10 Keywords Each
DepartmentExample Keywords
Apps teamoutlook, erp, adobe, 365, crashing
Break-Fix teamprojector, laptop, screen, keyboard
Deployment Teamipads, imaging, upgrade, monitors
Infrastructureeduroam, vpn, server, dns, wi-fi
Labs teamlab, spss, cad, toner, freeze
Macro teamsql, sync, script, database, hr
Online Learningd2l, zoom, respondus, panopto
Phone & Wiringvoip, ethernet, port, cable, phone
The bigger picture

The IT helpdesk router is one custom script — built in a short time for one specific department. The same Eulyph operators (@, ? ->, env(), .serve()) can power entirely different workflows for entirely different teams. That's the point: Eulyph is a general-purpose language whose AI-native syntax makes it adaptable to any domain without starting from scratch.


// roadmap — what eulyph can power
Built for any industry. Ready to adapt.
Because Eulyph routes by meaning rather than hard-coded rules, the same language works across entirely different domains. Below are active work and planned future applications.
● Active
📬

IT Helpdesk Automation

Autonomous ticket routing across 8 departments with confidence scoring, AI reasoning, and JSON export. Already demonstrated live.

○ Planned
🏦

Finance & Auditing

Watch financial feeds, extract figures from messy reports, and use ? -> to autonomously flag suspicious transactions for human review — no hard-coded rules needed.

○ Planned
🏠

Real Estate & Business

Extract prices, dates, and terms from unstructured property listings or contracts. Route inquiries, generate summaries, and automate follow-ups without fragile regex patterns.

□ Future
⚕️

Healthcare (std.medical)

A planned domain-specific standard library with pre-confirmed AI models trained on medical data — enabling intent-driven triage, documentation, and records routing.

□ Future
⚖️

Legal (std.legal)

Contract analysis, clause extraction, and obligation flagging — without building a specialized NLP pipeline. Eulyph reads meaning, not just keywords, across any legal document format.

□ Future
🚀

Multi-language Compilation

Prototype strategy in Eulyph, then compile the performance-critical sections into C++ or Rust — combining AI-native flexibility with systems-level speed.


// licensing
Open for learning. Licensed for business.
Eulyph is protected under a custom Non-Commercial License governed by the laws of the State of Minnesota. Use it freely for personal projects, education, and research — commercial use requires a separate agreement.

🔓 Non-Commercial — Free

Personal projects, education, academic research, testing, and non-profit use are covered at no cost. Distribute freely under the same license terms with proper attribution.

💼 Commercial — License Required

Any use that generates revenue, reduces costs in a business, or delivers Eulyph as part of a product or SaaS offering requires a separate written commercial license from Eulyph LLC.

📍 Governing Law

Governed by the laws of the State of Minnesota, USA. Disputes resolved in Minnesota courts. Includes standard no-warranty and limitation-of-liability provisions.


// faq
Common questions
Why build a new language instead of just adding AI to Python?
Eulyph treats AI as a native part of the language — not a library call. The @ and ? -> operators are first-class syntax, which means developers write fluid intent rather than mechanical steps. No Python add-on can replicate a language where the grammar itself understands what you mean.
Does Eulyph send my data to the cloud?
No. Eulyph connects to a local LLM endpoint (like LM Studio at localhost:1234) by default — your data stays on your machine. You can configure it to point to any compatible endpoint, but there's no cloud dependency built into the language itself.
Is the IT helpdesk script part of Eulyph's core download?
No — live_email_dispatch.eulyph is a custom script built specifically for an IT department demonstration. It's a great example of what you can build with Eulyph, but the core language download is the interpreter itself (the 8 Python files). You write your own .eulyph scripts on top of it for whatever domain you need.
Can Eulyph work with existing Python libraries?
Yes. The extern keyword imports any Python module directly into Eulyph. The env() sandbox lets you run arbitrary Python via py.exec() and retrieve values with py.eval(). Eulyph has full access to Python's ecosystem without sacrificing its own clean syntax.
How do I run a .eulyph file?
Clone the repo from GitHub, install Python 3.10+, Flask, and requests, start your local LLM endpoint, and run python main.py from the project root. Point it at your .eulyph file and you're live. Full setup instructions are in the GitHub repo.

// the team
Built by
IH
Isaiah Huus
JZ
James Zook
AI
Abid Imtiaz
🔒 Eulyph LLC · Non-Commercial License · All commercial rights reserved

Ready to build with Eulyph?

View the source, explore the codebase, and download the package — all on GitHub.

View on GitHub