About Injoys
Injoys is not a site that merely displays articles. It is a content data publishing service that distributes content as data packages.
We publish only content produced by our editors together with AI under field-specific editorial standards, checked against verified sources. We do not host user submissions.
All content is provided in Korean and English by default, with Japanese and Spanish added, and the content shown to people always matches the data provided to AI and search engines.
Our editors decide the topics and editorial standards, AI produces the draft, translations and images under those standards, and before publishing we check the figures against source material and withhold anything that falls short of our review criteria. The result is published simultaneously as a human-readable page and a machine-friendly file set.
Our goal is to supply trustworthy structured knowledge to the global search and AI ecosystem.
How we make our content
Instead of claiming quality, we publish the steps every article goes through. They apply to every article (dad-joke articles - humour written for this site - have no factual sources, so they skip the source and figure-check steps), and you can see the result on each article page.
- Our editors decide the topics and the editorial standards for each field. Where accuracy matters most - subsidies, taxes, insurance, legal procedures - we limit the scope to rules, procedures and published data, and rule out product recommendations or investment and medical advice.
- A draft is produced under those standards, and the reference material behind it is published alongside as sources.
- Figures such as amounts, periods and dates are checked against the source material before publication. The number of corrections and the timestamp appear in the Verification section of every article, and the corrected values themselves are listed on the original-language page.
- Articles that fall short of our review score or the image requirement are not published. Translations go through cross-verification too, and languages that fall short stay unpublished.
- Unless a source is credited, article images are made with AI. Newly made images carry an AI-generated marker inside the file (IPTC metadata), and those used in articles from fields where an image could pass for a real photo (Culture & entertainment, Policy & benefits, History & resilience) and in retold tales and humor (Folk tales, Dad jokes) also show an “AI-generated image” label on screen. All audio is AI-synthesized, and the player shows “AI-generated voice”. External links in the body are checked daily and taken down automatically once they stop resolving.
Principles
- Only content our editors define and oversee is published - no user submissions.
- Every article that deals with facts ships with the reference material it drew on, and figures in the body are checked against source material before publication.
- The same content for people and machines - no cloaking.
- Every article ships with at least two relevant images with alt text.
- Multilingual versions are localized, not merely translated.
Listen to content
You can listen to published articles as AI speech. The audio is pre-generated on the server and plays instantly on most devices, including mobile, free with no login required.
Explore our features →
Available languages
Content is served in two tiers depending on review status. Reviewed tier 1 languages get everything from human-readable pages to audio and embeddings, while machine-translated tier 2 languages are served only as data files and through the API. The reason for the split is simple: we do not publish unreviewed translations as pages for people.
| Language |
Value (lang) |
Tier |
Web pages |
Data files |
Audio (TTS) |
Embeddings |
|
한국어
|
ko |
Tier 1
|
✓ |
✓ |
✓ |
✓ |
|
English
|
en |
Tier 1
|
✓ |
✓ |
✓ |
✓ |
|
日本語
|
ja |
Tier 1
|
✓ |
✓ |
✓ |
✓ |
|
Español
|
es |
Tier 1
|
✓ |
✓ |
✓ |
✓ |
|
Indonesian
|
id |
Tier 2
|
- |
✓ |
- |
- |
|
Portuguese
|
pt |
Tier 2
|
- |
✓ |
- |
- |
|
Chinese (Traditional)
|
zh-hant |
Tier 2
|
- |
✓ |
- |
- |
|
Deutsch
|
de |
Tier 2
|
- |
✓ |
- |
- |
- Tier 1 - Languages whose translations have been reviewed. Everything is provided: human-readable web pages, search exposure (sitemap and hreflang), AI audio (TTS) and embeddings.
- Tier 2 - Machine-translated, data-only languages. Served through data files and the API only - no web pages, no search exposure, no audio (TTS) and no embeddings. Every file carries a translation_status: machine label to mark that it has not been reviewed.
Datasets
Beyond human-readable pages and machine-readable data files, content is also served as datasets ready for model training, fine-tuning and RAG indexing. There are four kinds - a text corpus, instruction data built from FAQs, a parallel corpus aligning one article across languages, and embedding vectors - and the translation review state is always labeled.
Bulk datasets are served through the paid API, while the unauthenticated data files that serve one article at a time remain free exactly as they are today.
About the dataset API
Data Access ·
License
Injoys