AI Disclosure

BEATSYNC PRO

BeatSync PRO — AI Usage

BeatSync PRO uses AI in two distinct ways:

1. Proprietary AI Vision — Clip Analysis

When you add video clips to your project, BeatSync PRO can analyze each clip using proprietary AI vision to understand its visual content — scene type, motion patterns, color palette, mood, and energy level.

• How it works: Individual frames are sent to a secure API endpoint for analysis
• What is sent: A single video frame (image data only). No personal information, filenames, or metadata is attached
• What is returned: A structured scene analysis (JSON) describing visual properties of the frame
• Data retention: Frames are processed in real-time and are not stored on any external server
• Caching: Analysis results are cached locally on your machine. Once analyzed, a clip never needs to be re-analyzed — even across projects
• Optional: AI clip analysis is not required. You can manually assign clip properties or skip analysis entirely

2. 15-Agent AI Production System — Local

BeatSync PRO's 15 AI agents operate as an internal creative coordination system:

• What they do: Audio analysis, beat detection, effect coordination, clip selection, quality control, creative direction, and production planning
• Where they run: Entirely on your local machine. The agent system uses a local message bus, local knowledge graph, and local decision engine
• What they access: Your audio file (for beat/frequency analysis), your video clips (for matching), and cached analysis data
• What they produce: A production plan — which clips to use, which effects to apply, and how to synchronize everything to the beat
• No external calls: Agent decisions, production meetings, quality gates, reward calculations, and intensity adjustments are all computed locally

What AI Does NOT Do

• AI does not generate your video frames — all rendering uses your GPU via GLSL shaders and NVENC encoding
• AI does not generate your music or audio
• AI does not upload your finished videos anywhere
• AI does not access your filesystem beyond the clips and audio you provide
• AI does not run in the background when BeatSync PRO is closed

CLAREON

Clareon — AI Usage

Clareon uses a proprietary neural network called ClareonNet for video upscaling and enhancement, plus an optional AI Agent Intelligence system for enhanced processing.

1. ClareonNet — Local Neural Network

• Architecture: ClareonNet is a custom 2.08M parameter neural network with SpatialEncoder, bidirectional temporal processing, and encoder skip connections. Trained 100 epochs from scratch specifically for video upscaling.
• Where it runs: 100% local. ClareonNet runs entirely on your GPU via PyTorch CUDA with FP16 inference. No frames, video data, or processing results ever leave your machine.
• Additional local models: CodeFormer (face enhancement) and AI optical flow (frame interpolation) also run entirely on your local GPU.
• Encoding: NVENC hardware encoding (H.264, H.265, AV1) with automatic CPU fallback.
• Training data: ClareonNet was trained on licensed and original video data. No user data is used for training.

2. AI Agent Intelligence — Optional, Uses Credits

When enabled by the user, 5 AI agents enhance the upscaling pipeline:

• LENS (Vision Analyzer) — Analyzes frame content and scene characteristics
• PRISM (Strategy Director) — Selects optimal enhancement strategy per video
• FORGE (Enhancement Agent) — Guides frame-level enhancement decisions
• TEMPORAL (Consistency Agent) — Ensures temporal coherence across frames
• REFINE (Quality Evaluator) — Evaluates output quality and adjusts parameters

• How it works: Agent analysis requests are sent to a secure API endpoint. Processing decisions are returned and applied locally.
• Optional: AI Agent Intelligence is toggled on/off in the UI. When off, Clareon operates 100% locally with no external connections.
• Credits: Agent-enhanced processing uses AI credits from your account balance.

What AI Does NOT Do

• AI does not generate new content — it enhances and upscales existing video frames
• AI does not hallucinate or fabricate visual details — the skip connection architecture preserves source accuracy
• Core upscaling (ClareonNet) never sends data externally — only optional AI agents connect to the API
• AI agents are completely optional and disabled by default

PROMETHEUS SHIELD

Prometheus Shield — AI Usage

Prometheus Shield is a deterministic code protection system built on Abstract Syntax Tree (AST) transformations, classical cryptography, and compiler technology. Its 17 protection layers operate through:

• AST transformations: Python source code is parsed into an abstract syntax tree and transformed through 12 standard + 5 enterprise layers of obfuscation
• Classical & post-quantum cryptography: AES-256-GCM, HMAC-SHA512, ML-KEM-768 (FIPS 203), ML-DSA-65 (FIPS 204)
• Native compilation: Nuitka compiles transformed Python into C, then into a standalone native executable
• Runtime protection: Anti-debug detection, anti-tamper verification, and self-healing code use deterministic algorithms, not machine learning

No neural networks, no training data, no probabilistic models. Every transformation is mathematically defined and reproducible.