Overview

RAWSUD Artist Intelligence helps moderators review artist applications faster, more consistently, and with full human oversight.

RAWSUD Music is an independent music platform for artists, fans, streaming, merchandise, and editorial culture. During OpenAI Build Week 2026, RAWSUD extended its existing artist onboarding flow with Artist Intelligence: a hybrid review pipeline that combines deterministic validation rules with GPT-5.6 Luna analysis.

The system reviews artist applications and existing artist profiles, generates structured findings, produces actionable recommendations, and preserves review history. It is designed to support artists across genres, including hip hop, rock, pop, electronic music, DJs, producers, bands, crews, and other independent music scenes.

Artist Intelligence is not an automatic approval system.

OpenAI assists the review process, but the final approval, rejection, or request for changes always belongs to a human administrator.


The Problem

Manual artist reviews become slow, repetitive, and difficult to standardize as a platform grows.

Before Artist Intelligence, moderators manually inspected artist applications and repeated the same quality checks for every profile. As the number of artists increased, maintaining consistent review quality became increasingly difficult.


The Solution

RAWSUD Artist Intelligence adds a structured review workflow to the existing onboarding system.

Review workflow

Artist Submission

Data Validation

Deterministic Rules

GPT-5.6 Luna

Findings & Recommendations

Human Decision

Audit History

The system includes:

  • Artist application review workflow
  • Existing artist informational review
  • Deterministic validation engine
  • Versioned rule snapshots
  • GPT-5.6 Luna analysis behind a feature flag
  • Structured findings and recommendations
  • Approve / Reject / Request Changes actions
  • Review history and audit trail
  • Admin and moderator review interface
  • Applicant-safe correction workflow
  • Responsive admin and artist interfaces

When OpenAI is disabled or unavailable, deterministic validation still generates a complete review report.


How We Used OpenAI

RAWSUD demonstrates both sides of the OpenAI ecosystem.

OpenAI development tools helped build the project, while OpenAI models power part of the product itself.

  • ChatGPT supported product reasoning, architecture discussions, acceptance criteria, and documentation.
  • Codex supported implementation, debugging, repository validation, documentation, and release preparation.
  • GPT-5.6 Luna is integrated as an optional review layer that generates structured findings and recommendations.

The OpenAI stage is protected by feature flags and environment variables.

Artist content is treated as untrusted input.

GPT-5.6 Luna never:

  • approves artists
  • rejects artists
  • changes user roles
  • publishes profiles
  • touches payments
  • changes artist visibility

AI assists. Humans decide.


Real OpenAI Execution

A controlled production execution was completed for Build Week.

Field Value
Run #8
Target EXISTING_ARTIST
QA Profile Lunar Norte
State COMPLETED
AI Status AI_COMPLETED
Model GPT-5.6 Luna
Duration 4946 ms
Total Tokens 998
Prompt Tokens 552
Completion Tokens 446
Reasoning Tokens 63
Findings 5
Recommendations 5

Lunar Norte was used only as the QA evidence profile.


Why GPT-5.6 Luna

GPT-5.6 Luna was selected because this workflow requires:

  • structured JSON outputs
  • deterministic moderation support
  • latency-sensitive analysis
  • cost-conscious execution
  • actionable recommendations

The administrator always keeps the final decision.


How It Was Built

Artist Intelligence extends the existing artist_applications workflow instead of creating a separate system.

Implementation includes:

  • Next.js frontend
  • React
  • Express backend
  • MySQL persistence
  • Deterministic rule engine
  • Optional GPT-5.6 Luna review service
  • Node.js automated tests

Supported review targets:

  • ARTIST_APPLICATION
  • EXISTING_ARTIST

Existing artist reviews are informational only and never modify artist roles, visibility, streaming, merchandise, or payment systems.


Safety & Responsible Design

The project is built around human oversight.

Safeguards include:

  • Deterministic validation before AI
  • Feature-flag controlled OpenAI integration
  • Strict JSON validation
  • Human approval authority
  • Persistent audit logs
  • Stored review history
  • Untrusted-input handling
  • No automatic publication or rejection
  • No interaction with payments, royalties, merchandise or checkout

Validation

Validation performed:

  • node --check backend/server.js
  • node --test backend/tests/artist-review/*.test.js
  • npm run lint
  • npm run build
  • npm audit --prefix backend --audit-level=high

Results:

  • Backend syntax ✔
  • Backend tests ✔ (9 passed)
  • Frontend lint ✔
  • Frontend build ✔
  • Dependency audit reviewed ✔

The public repository contains no secrets, SQL dumps, uploads or private environment files.


What We're Proud Of

This is not a prompt wrapped inside a UI.

Artist Intelligence is integrated into an existing production architecture with:

  • persisted review history
  • rule snapshots
  • human decision workflows
  • applicant-safe status views
  • deterministic fallback
  • documented QA
  • real GPT-5.6 Luna execution
  • clean public repository

The platform owner always remains in control while OpenAI helps make moderation faster, clearer, and more consistent.


Challenges

The biggest challenge was integrating AI without creating an unsafe approval system.

OpenAI improves the review process while remaining isolated from payments, authentication, streaming, merchandise, checkout, and publication workflows.

Another challenge was making every review auditable through persisted rules, findings, recommendations, and decisions.


What's Next

Next planned improvements:

  • Disposable MySQL integration testing
  • Full staging validation
  • Additional deterministic review rules
  • Richer moderator analytics
  • Continued dependency maintenance
  • Expanded Artist Intelligence capabilities while preserving human approval

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for RAWSUD Artist Intelligence

Built With

Share this project:

Updates