Knowledge base›How we work
Our Roadmap
Transparency is fundamental to how we analyze books. Here's what's available now and what's coming next.
How Subtext Works
Subtext analyses publicly available book descriptions and metadata to identify common sensitive themes (such as violence, abuse, or discrimination).
It uses automated language analysis to surface patterns and provides severity and age guidance to help readers make informed choices.
Subtext does not store personal reading histories or profiles.
Phase 1: Available Now
Automated Content Warnings
Detailed, automated alerts for sensitive topics with severity ratings (mild, moderate, severe) and comprehensive reasoning. Every warning includes detailed explanations of how it was determined.
Age Ratings
Australian Classification Board-based age ratings (G, PG, M, MA15+, R18+) with clear age recommendations and methodology explanations.
Thematic Analysis
Deep analysis of themes, tone, and emotional weight. Understand what a book explores beyond just content warnings.
Multi-Source Metadata
Book information gathered from multiple trusted sources (Google Books, Open Library) with web search enrichment when descriptions are limited.
Phase 2: Coming Next
Source Citations
Coming SoonEvery warning will include source URLs so you can verify exactly where information came from—author sites, publisher pages, book databases, or community sources.
Author-Verified Warnings
Coming SoonWhen authors provide their own content notes, we'll prioritize those above all other sources. Author-verified warnings will be clearly marked and displayed first.
Community Feedback
Coming SoonRate warnings as helpful or not helpful. Community feedback will help refine accuracy over time and surface the most reliable information.
Our Principles
Transparency First
Automated warnings include detailed reasoning so you can understand how each warning was determined. Source URLs coming in Phase 2.
Author Authority
When authors provide their own content notes (Phase 2), we'll prioritize those above all other sources. They know their work best.
Multiple Verification
We gather information from multiple sources (Google Books, Open Library, book databases). Community feedback in Phase 2 will help refine accuracy over time.
Subjective Severity
Severity ratings are subjective—they vary by individual sensitivity and the nature of the content. We provide guidance, not absolutes.
User Privacy
We don't sell your reading data. Your scans are anonymous, and we don't train models on individual user preferences.
What You Can Verify Now
For every book, you can currently see:
- The system's detailed reasoning for each severity rating
- How severity is calculated based on frequency, explicitness, proximity, and context
- Age ratings with methodology explanations
- Thematic analysis and emotional weight assessments