AI Model License Checker

• CLicense is the closest lightweight competitor, covering Hugging Face, GitHub, and public AI repositories.
• SciSpace Dataset License Checker accepts a license, URL, or text plus intended use, but focuses on one-license Q&A and research compliance.
• [Data Provenance Initiative](https://www.dataprovenance.org/) audits origins, licenses, creators, and derivation chains for public datasets. It is a research/data asset, not a small-team pre-launch workflow.
• ProvenanceAI, Relyance AI, and similar vendors address broader provenance, governance, and enterprise compliance with heavier scope and sales motion.
• [SPDX's AI work](https://spdx.dev/learn/areas-of-interest/ai/) shows that AI-SBOM/AI BOM is emerging as a standardization direction and can guide export fields.
• **SERP occupancy (observed, not rank-tested):** dataset license checker already returns direct tools and research assets; offline transcription software is dominated by commercial apps and roundups. A localized incognito Top 3/Top 10 audit and DataForSEO validation remain undone, so ranking ease cannot be claimed.
• **Entry wedge:** start from one Hugging Face model URL, use intended use to trace model → base model → declared datasets, preserve source evidence for every finding, and export an AI BOM that clients or counsel can review. The product must not claim a legal conclusion.

MVP Checklist

1. 1.Define the report schema, component relationships, findings, evidence, intended-use profile, and rule version.
2. 2.Implement Hugging Face URL parsing, Hub API wrapper, rate-limit errors, and partial-success states.
3. 3.Implement one-level model → base model → datasets tracing with raw evidence retention.
4. 4.Implement license normalization and deterministic rules with tests for every finding.
5. 5.Build Scan, Report, dependency list, remediation checklist, and JSON/CSV/Markdown exports.
6. 6.Add Home, Methodology, About, FAQ, Privacy, and fixed public examples.
7. 7.Add `$49 Audit Pack`, `$199 pilot`, Payment Link/priced form, and privacy-safe events.
8. 8.Complete mobile, accessibility, SEO, caching, rate limits, production build, and regression testing.

Don't Build

• ✕Do not expand features without approval.
• ✕Do not add a complex backend; V1 needs only controlled API routes and short-term caching.
• ✕Do not build custom auth, memberships, orders, subscription billing, or an admin backend first; keep lightweight commercial validation.
• ✕Do not sacrifice launch speed for superficial completeness.
• ✕**Do not cut live Hub API retrieval, dependency tracing, evidence, deterministic rules, or AI BOM export to make the product “lighter.”**
• ✕Do not use an LLM to generate legal conclusions or treat missing metadata as infringement.
• ✕Do not label a green state “legal/approved,” claim to replace counsel, or collect private Hugging Face tokens.