⚡ VOLT-AI

Intelligent Multi-Model AI Coordination, Routing & Observability Engine

---

⚠️ Intellectual Property & Licensing Notice

The VOLT-AI Model (volt-ai-1.0 specifications, weights, and inference schema) is released as Open Source under the Apache License 2.0.

The VOLT-AI Platform, Core Coordination Engine, Dynamic Multi-Model Router, Secret Scrubbing Pipeline, and Observability Backend are proprietary intellectual property created by S Sanjai (Baizo). All platform rights are strictly reserved. Commercial redistribution, unauthorized resale, or hosting of competing SaaS proxy services without express prior written permission is prohibited. See LICENSE for full legal terms and credits.

---

🌟 Overview

VOLT-AI provides next-generation AI coordination, dynamic specialist fallback, real-time observability, and a community model portal:

• Free Starter Tier (Zero Setup Required): Anyone can start coding immediately with VOLT-AI without an API key (up to 30 free daily requests).
• Web Developer Portal & 1-Click API Key Generator: Visit volt-ai-lmgz.onrender.com to generate your personal volt_live_... API key with 50 free requests per day, copy-paste snippets for Python, Cursor, VS Code, and terminal CLI.
• Bring-Your-Own-Key (BYOK) for Unlimited Power: Provide your personal provider key (OpenRouter, DeepSeek, Gemini, or OpenAI) for 100% unlimited requests billed to your provider at zero markup.
• Open-Source VOLT-AI Model (volt-ai-1.0): Open model card and inference schema for code generation, architectural synthesis, and multi-turn refactoring.
• Automated Data Collection Flywheel: All coding activity is scrubbed of credentials and continuously captured into PostgreSQL for fine-tuning future open-source VOLT releases via LoRA and DPO.
• Client Tooling (volt-coder & volt): Local terminal CLI for one-shot coding, interactive chat sessions, and personal token usage analytics.
• Universal Drop-in Proxy: OpenAI & Anthropic API compatible proxy supporting Cursor, Aider, OpenCode, VS Code extensions, and native SDKs.
• Zero-Leak Secret Redaction: In-flight redaction of API keys, AWS credentials, GitHub tokens, and private keys before persistence.
• Live Observability Dashboard: Real-time glassmorphism analytics dashboard tracking tokens, latency, cost, and test pass rates.

---

🔑 3 Ways to Access VOLT-AI

Tier	API Key Required?	Quota / Limits	Best For
1. Free Starter	❌ None	30 free requests / day	Instant testing, quick CLI queries, casual terminal scripting
2. Portal Developer Key	⚡ volt_live_... (1-Click Generate)	50 free requests / day	Personal developer projects, VS Code / Cursor plugins, scripts
3. BYOK (Bring Your Own Key)	🔑 OpenRouter, DeepSeek, Gemini, OpenAI	♾️ Unlimited	Power users, large codebases, continuous development

---

🔄 The BYOK Data Collection Flywheel

VOLT-AI connects model access directly with an automated data collection and fine-tuning flywheel:

┌─────────────────────────────────┐
│   Developer / External User     │
│   (Terminal CLI, Cursor, etc.)  │
└────────────────┬────────────────┘
                 │
                 │ 1. Supplies THEIR OWN provider API key (OpenRouter, DeepSeek, Gemini, OpenAI)
                 │    via 'volt activate' or 'Authorization: Bearer <your-key>'
                 ▼
┌────────────────────────────────────────────────────────────────────────┐
│                        ⚡ VOLT-AI Engine                               │
│                                                                        │
│ 2. BYOK Gatekeeper: Verifies the caller provided an active key.       │
│    (Rejects unauthorized requests with 401).                           │
│                                                                        │
│ 3. Execution: Dispatches request to provider using THE USER'S KEY.     │
│    (The provider bills the user's account — zero cost to host).        │
│                                                                        │
│ 4. Zero-Leak Redaction: Strips API keys, passwords, and tokens.        │
│                                                                        │
│ 5. Data Collection: Records prompts, generated code, latency, and     │
│    session metrics to PostgreSQL (llm_calls table).                    │
└────────────────┬───────────────────────────────────────────────────────┘
                 │
                 ▼
┌────────────────────────────────────────────────────────────────────────┐
│                   Continuous Learning Pipeline                         │
│                                                                        │
│ • POST /admin/export-dataset: Automatically exports SFT and DPO pairs. │
│ • training/train_lora.py: Fine-tunes future open-weight VOLT models   │
│   directly on real-world developer interactions!                       │
└────────────────────────────────────────────────────────────────────────┘

---

🤖 The VOLT-AI Model (volt-ai-1.0)

The VOLT-AI Model is released under the Apache 2.0 Open-Source License.

Model Profile

• Model ID: VOLT-AI / volt-ai-1.0
• Architecture: Multi-Model Coordinated Specialist Mixture
• Specializations: Code synthesis, test-driven repair, system architecture, rapid debugging
• Access Model: Bring-Your-Own-Key (BYOK)
• Context Routing:
  • code track: Specialized code generation and bug fixes
  • general track: Multi-turn architectural explanation and reasoning
  • fast track: Low-latency autocompletions and edits
  • creative track: Novel ideation and documentation

---

🛠️ Local Setup Configuration (For Developers & Teams)

You can set up and run the VOLT-AI client locally on your workstation to interact with the VOLT-AI cloud server or your self-hosted instances.

1. Clone & Install Client Dependencies

Clone the repository and install the client package locally:

# Clone the repository
git clone https://github.com/baizo7/VOLT.git
cd VOLT

# Set up a Python virtual environment (Python 3.9+)
python -m venv .venv

# Activate the virtual environment
# On Linux/macOS:
source .venv/bin/activate
# On Windows (PowerShell):
.venv\Scripts\Activate.ps1

# Install local client package in editable mode
pip install -e .

2. Configure Your Client (Free Tier or BYOK)

Option A: Zero Configuration (Instant Free Starter Tier)

You do not need an API key to test VOLT-AI! Simply run:

volt-coder "Write a binary search algorithm in Python"

Requests will automatically connect to the free community starter tier (up to 30 requests/day).

Option B: Generate a Free Developer Portal Key (50 requests/day)

1. Open the VOLT-AI Portal.
2. Switch to the 🔑 Model Portal & API Keys tab.
3. Click ⚡ Generate My Free VOLT API Key.
4. Run volt activate and paste your volt_live_... key.

Option C: Bring Your Own Provider Key (Unlimited Usage)

For unlimited requests, run:

volt activate --server https://volt-ai-lmgz.onrender.com

And provide your personal provider API key:

• OpenRouter (openrouter.ai/keys) (Recommended for access to all models)
• DeepSeek (platform.deepseek.com)
• Google Gemini (aistudio.google.com)
• OpenAI (platform.openai.com)

This saves your local configuration to ~/.volt-ai/config.json.

---

💻 Local CLI Usage (volt-coder & volt)

Once installed, use the local commands to generate code, chat, and inspect routing:

One-Shot Code Generation

# Basic prompt
volt-coder "Write a fast thread-safe LRU cache in Python"

# Include existing files in context
volt-coder code -f app/security.py "Review this file for security vulnerabilities"

Interactive Multi-Turn Terminal Chat

volt-coder chat

Check Available Specialists & Models

volt-coder models

Personal Usage Analytics

# View personal token consumption and latency
volt usage

# Open the real-time web dashboard
volt dashboard

---

🔌 IDE & Tool Integration

Route your existing developer tools and IDEs through VOLT-AI:

1. Cursor / VS Code / Aider / OpenCode (OpenAI SDK Compatible)

Add the following configuration to your terminal profile or IDE settings:

Windows PowerShell ($PROFILE):

$env:OPENAI_BASE_URL = "https://volt-ai-lmgz.onrender.com/v1"
$env:OPENAI_API_KEY  = "your-provider-api-key"   # e.g. OpenRouter, DeepSeek, or OpenAI key
$env:X_SESSION_ID    = $env:USERNAME

Linux / macOS (~/.bashrc or ~/.zshrc):

export OPENAI_BASE_URL="https://volt-ai-lmgz.onrender.com/v1"
export OPENAI_API_KEY="your-provider-api-key"    # e.g. OpenRouter, DeepSeek, or OpenAI key
export X_SESSION_ID="$USER"

2. Anthropic SDK & Claude CLI

export ANTHROPIC_BASE_URL="https://volt-ai-lmgz.onrender.com"
export ANTHROPIC_API_KEY="your-anthropic-key"

3. Model Name to Use in Tools

Specify either:

• VOLT-AI (Primary recommended coordinated model)
• volt-ai-1.0

---

📊 Live Observability & Cloud Service

• Central Server: https://volt-ai-lmgz.onrender.com
• Google Sign-In & Login Portal: https://volt-ai-lmgz.onrender.com/login
• Interactive Dashboard: https://volt-ai-lmgz.onrender.com/dashboard
• Health Check: https://volt-ai-lmgz.onrender.com/health
• API Models Catalog: https://volt-ai-lmgz.onrender.com/v1/models

---

🧪 Running Client Tests

Verify your local client installation:

# Run unit & API test suite (161 tests passing)
pytest -m "not integration"

---

⚖️ License, Copyright & Credits

Author & Credits

• Creator & Lead Architect: S Sanjai (Baizo)
• GitHub: @baizo7
• Repository: https://github.com/baizo7/VOLT

Dual-License Terms

1. VOLT-AI Model (volt-ai-1.0): Released under the Apache License 2.0. You are free to run inference, fine-tune, and build research applications on top of the model.
2. VOLT-AI Platform & Backend: Proprietary & Source-Available. Copyright (c) 2026 S Sanjai (Baizo). All rights reserved. The backend proxy architecture, coordination engine, database schemas, and dashboards are protected and cannot be redistributed commercially or hosted as a competing service without authorization.

Mandatory Attribution

Any application, research paper, derivative model, or service utilizing VOLT-AI must visibly credit:

"Powered by VOLT-AI — Created by Baizo (S Sanjai) | https://github.com/baizo7/VOLT"

For licensing inquiries or commercial enterprise deployments, please contact the author via GitHub.