1. Summary
RÊVE NATION is a release workflow platform. Some features use artificial intelligence — specifically AI writing assistance, audio feature extraction, and pitch structuring. This page explains which features use AI, which do not, and how AI assistance works in practice.
2. Which Features Use AI
- —
Artist bio and song description writing assist — AI suggests draft text based on your track data and profile. You review, edit, and save each suggestion yourself.
- —
Mood classification — Audio features extracted from your track are classified against a fixed vocabulary of mood tags (e.g., Upbeat, Melancholic, Aggressive). Output is one or more tags from a bounded set — not open-ended generation.
- —
Pitch structuring assistance — When you use a DSP Pitch Builder, AI helps structure the pitch copy in the format that platform expects. The content is based only on facts you provide about your track and artist profile.
- —
Release debrief narrative — A written summary of your track's audio characteristics and release positioning. Presented for review before you use it.
3. Which Features Do NOT Use AI
- —Release timeline calculation (milestone dates are computed from your release date, not generated)
- —Deadline tracking and notifications
- —File processing and audio feature extraction (librosa + ffmpeg — deterministic, not AI)
- —Billing, subscriptions, and payments
- —Knowledge Hub content (written by humans, not AI-generated)
- —Artist profile data entry and storage
4. AI Never Acts Autonomously
Every AI output requires your review before anything happens.
- —AI writing suggestions are presented as drafts — you choose to save, edit, or discard each one.
- —Pitch copy is presented in the builder — you copy and submit it yourself through your own DSP accounts.
- —The AI never sends, posts, submits, or publishes anything on your behalf.
- —No AI output is automatically applied to your profile or release without your explicit action.
5. Which AI Provider Is Used
We use Anthropic's Claude API for writing assistance features. Audio feature extraction uses librosa (open-source) and ffmpeg — neither of which involve external AI providers.
What we transmit to Anthropic: track metadata (title, genre, BPM, key, audio features), and the text context needed to generate the specific output requested (bio drafts, pitch copy, debrief narratives).
What we do NOT transmit to Anthropic: your raw audio files, your email address, your name, or any payment information. Anthropic processes this data subject to their privacy policy (anthropic.com/legal/privacy) and is not permitted to use your data to train their models under our API agreement.
6. Accuracy and Limitations
- —
AI-generated copy may contain inaccuracies. Review pitch copy, bios, and narratives critically before use. Do not submit AI-generated text to DSPs without reviewing it for accuracy.
- —
No guarantee of commercial outcomes. AI assistance improves pitch quality but does not guarantee editorial placement, streaming performance, or label interest.
- —
Statistical variation. The same inputs may produce different text outputs across runs. This is normal and expected for generative AI.
7. Your Responsibility
RÊVE NATION is a tool to inform and support your release decisions — not to make those decisions for you. You are responsible for:
- —Reviewing all AI-generated content before using it.
- —Verifying factual claims in AI-generated pitch copy before submitting to DSPs.
- —Making your own editorial, creative, and business decisions.
- —Ensuring any content you create based on our outputs complies with applicable laws and third-party rights.
Questions about AI use in the Service? support@revenation.cc