Over the past year, I have been building something I am genuinely excited about: a suite of AI-powered platforms that help people learn languages by reading real news. The three sites — MotsActu for French, PrensaFacil for Spanish, and YomuDaily for Japanese — each take current events and transform them into graded reading material for language learners.
Here is how they work and what I learned building them.
The Problem with Traditional Language Learning
Most language learning tools give you artificial, contrived content. “The cat is on the table.” “Where is the library?” These sentences teach grammar, but they do not prepare you for actually reading in your target language. Real-world content — news articles, opinion pieces, cultural commentary — uses vocabulary, idioms, and structures that textbooks rarely cover.
But real news articles are overwhelming for intermediate learners. The vocabulary is too advanced, sentences are too complex, and there is no scaffolding to help you understand what you are reading.
That is the gap these platforms fill.
How AI Powers the Content Pipeline
Each platform follows the same core workflow:
- Source selection. Current news stories are identified based on topic relevance, cultural significance, and pedagogical value.
- AI-assisted adaptation. Stories are rewritten at multiple reading levels, preserving the core information while adjusting vocabulary complexity, sentence length, and grammatical structures.
- Vocabulary scaffolding. Key terms are highlighted with definitions, usage notes, and example sentences. Reading level badges help learners find content at their proficiency level.
- SEO and distribution. Each story is optimized for search, submitted to Google for indexing, and distributed through newsletters and social media.
The entire pipeline — from story selection to published article — runs with minimal manual intervention. AI handles the heavy lifting of content adaptation, while I focus on quality control, platform development, and growth strategy.
The Tech Behind the Platforms
All three sites run on WordPress with custom classic themes, hosted on DreamHost. The content pipeline integrates with the WordPress REST API for publishing, the Google Search Console API for SEO monitoring, and MailerLite for newsletter distribution.
Deployment is automated through GitHub Actions with SFTP, so pushing code to the main branch deploys to production immediately. Each site has its own repository, theme, and configuration, but they share the same architectural patterns.
The AI components use Claude for content generation and adaptation, with custom prompts tailored to each language and reading level. The quality of AI-generated educational content has reached a point where, with proper prompting and review, it is genuinely useful for learners.
What I Have Learned
Building these platforms has reinforced a few principles:
- AI is a force multiplier, not a replacement. The platforms would not exist without AI — the content volume would be impossible to sustain manually. But AI alone does not produce great educational content. The system design, quality standards, and pedagogical framework are what make it work.
- WordPress is still underrated. For content-heavy sites with regular publishing schedules, WordPress with a custom theme is hard to beat. The ecosystem, the REST API, and the block editor provide everything these platforms need.
- SEO is a long game. Growing organic traffic to niche educational sites takes months of consistent publishing, technical optimization, and patience. There are no shortcuts.
What Is Next
The platforms are growing steadily. The focus now is on expanding the content library, improving the reading experience, and building community features that keep learners coming back. If you are learning French, Spanish, or Japanese, I would love for you to check them out.