Neo AI

Neo AI is a local AI assistant from Zenthralix-Lab, built with FastAPI, Ollama, and a browser dashboard. It runs on your own machine with local models for chat, system metrics, voice input, and basic local command execution.

Features

• Local Ollama chat with streaming responses.
• Default Neo models: neo-light:latest and neo:latest.
• FastAPI backend with REST and WebSocket endpoints.
• Browser dashboard with CPU, RAM, disk, network, and process metrics.
• Browser voice input through the Web Speech API.
• Bounded Neo ASI runtime context from /home/zenthralix/Documents/ASI NEO when available.
• Install and launch scripts for Linux, macOS, and Windows through WSL2.

Requirements

• Linux, macOS, or Windows with WSL2.
• Python 3.10 or newer.
• Ollama.
• curl or wget.
• 4 GB RAM minimum for neo-light:latest; 8 GB or more recommended.

Quick Start

git clone https://github.com/NeoOptimize/Neo-AI.git
cd Neo-AI
chmod +x install.sh
./install.sh

Open the dashboard:

http://localhost:8765

Manual Setup

curl https://ollama.com/install.sh | sh
ollama serve &

git clone https://github.com/NeoOptimize/Neo-AI.git
cd Neo-AI

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

ollama pull neo-light:latest
ollama pull neo:latest

cd neo-core
python3 neo_server.py

On Windows, use WSL2 and run the Linux setup inside the WSL distribution.

Models

Model	Purpose	Notes
neo-light:latest	Default fast chat	Best for low RAM systems
neo:latest	Full Neo model	Use when more RAM/CPU is available

Use the smaller model when the machine is under memory pressure:

export NEO_DEFAULT_MODEL=neo-light:latest

Configuration

Variable	Default	Description
NEO_PORT	8765	FastAPI server port
OLLAMA_BASE_URL	http://localhost:11434	Ollama API URL
NEO_DEFAULT_MODEL	neo-light:latest	Default chat model
NEO_CHAT_TIMEOUT	120	Chat request timeout in seconds
NEO_OLLAMA_KEEP_ALIVE	5m	Ollama model keep-alive value
NEO_OLLAMA_NUM_CTX	2048	Ollama context window
NEO_ASI_CONTEXT	/home/zenthralix/Documents/ASI NEO	Optional Neo ASI context source
NEO_ASI_CONTEXT_ENABLED	1	Set to 0 to disable runtime context injection

API

Base URL:

http://localhost:8765

Key endpoints:

GET  /health
GET  /api/status
GET  /api/models
GET  /api/metrics
GET  /api/neo/context
POST /api/chat
WS   /ws/chat
WS   /ws/monitor

Example non-streaming chat:

curl -s http://localhost:8765/api/chat \
  -H 'Content-Type: application/json' \
  -d '{"message":"Reply with OK","model":"neo-light:latest","max_tokens":16}'

Project Structure

neo-ai/
├── neo-core/
│   ├── neo_server.py
│   ├── neo_dashboard.html
│   └── neo_context.py
├── scripts/
│   ├── setup-models.sh
│   └── check-models.sh
├── .ai/skills/neo/
│   └── asi_context.md
├── config.py
├── config_v2.py
├── install.sh
├── requirements.txt
└── README.md

Troubleshooting

Ollama is not running

ollama serve
curl http://localhost:11434/api/version

Model not found

ollama pull neo-light:latest
ollama pull neo:latest
ollama list

Slow responses or high RAM

• Use neo-light:latest.
• Close heavy applications before loading neo:latest.
• Lower context size with NEO_OLLAMA_NUM_CTX=1024.
• Stop loaded models when done:

ollama ps
ollama stop neo:latest
ollama stop neo-light:latest

Voice input says no speech

• Use Chrome, Edge, or Safari.
• Allow microphone permission for http://localhost:8765.
• Confirm the microphone works in the OS sound settings.
• Speak after the voice waveform appears.

Support Zenthralix-Lab

NeoOptimize is released as a free public utility. Support helps Zenthralix-Lab fund testing hardware, build infrastructure, documentation, security review, release maintenance, and future free software projects for the community.

Email: neooptimizeofficial@gmail.com

Buy Me a Coffee: https://buymeacoffee.com/nol.eight

Saweria: https://saweria.co/dtechtive

Dana: https://ik.imagekit.io/dtechtive/Dana

About

Made with love at Zenthralix-Lab with Codex.

License

This project is released under the MIT License. See LICENSE.