Universal Neural Intelligence (Uni) Engine

The Deterministic Intelligence Operating System

Patents Pending:

* TAPESTRY – Long-Term Memory Summarization System
* LSP – Language Structure Protocol for Deterministic Parsing

Requirements:

Make sure you have an .env file in the root of each phase with the GROQ_API_KEY value as GROQ_API_KEY=gsk_XXX

Introduction

The Universal Neural Intelligence (Uni) Engine is a modular, language-driven AI framework designed for real-time command processing and continuous learning. It combines deterministic parsing with optional LLM fallback and includes long-term memory via summarization. The system treats language as a formal protocol: it parses, interprets, stores, and acts on natural language commands with verifiable precision.

Architecture Overview

Input and Preprocessing

Supports text input (and optionally voice/sensors) and prepares data for parsing.

Flexible Language Engine (FLE)

• RegEx Command Layer
• Constituency Parsing
• Dependency Parsing

Protocol Layer

• Data Protocols (internal execution)
• Actuation Protocols (external control)

LLM Fallback & L.E.A.R.N.

Fallback to a large language model when deterministic parsing fails. Uses the L.E.A.R.N. loop to refine and store new patterns.

Long-Term Memory Store (LTMS)

Summarizes conversations and stores them in a SQLite database for recall and continuity.

Project Structure

Each phase is in a self-contained folder. Each folder includes its own Flask app, database, virtual environment (env_uni), and requirements.txt.

/
|-- uni-alpha
|-- uni-bravo
|-- uni-charlie
|-- uni-delta
|-- uni-echo
|-- uni-foxtrot
|-- README.md

Phase Overview

uni-alpha Basic LLM-powered chatbot. Stores conversations.

uni-bravo Adds summarization and memory. Conversations are compressed into summaries.

uni-charlie Adds regex command parsing. Known queries are handled without LLM.

uni-delta Adds intent detection and function protocol mapping. Introduces structured execution.

uni-echo Full syntactic parsing with Stanza. Handles complex sentences and structures.

uni-foxtrot Patterns and configuration are data-driven via the database. Major refactor and stability pass.

Running a Phase

1. Navigate to a phase folder (e.g., cd uni-charlie)
2. Activate the virtual environment:
   • Windows: env_uni\Scripts\activate
   • Mac/Linux: source env_uni/bin/activate
3. Install dependencies:
   • pip install -r requirements.txt
4. Run the Flask app:
   • python app.py
5. Visit http://localhost:5011 (the port in app.py)

Each phase runs independently. Shut down with Ctrl+C and switch folders to run a different phase.

Git Ignore

One .gitignore in the root covers everything. It should include patterns like:

**/env_uni/ *.pyc pycache/ *.db .env

Note: This project uses a PyTorch nightly build. To install correctly, follow PyTorch nightly instructions. The provided requirements.txt expects nightly wheels (e.g., torch==2.9.0.dev...). Use pip install --pre torch or download the specific wheel from the official site.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.