Being upfront about how this works matters — both because the audience deserves it and because hiding the AI part would be silly.
A high-school student runs the project end-to-end: picking the topics, writing the prompts, reviewing the AI output, designing the site, and shipping the code. Her dad pays for the tools (Claude Code, the API bill, the server) and acts as a sounding board when she hits something tricky. The editorial calls and the build itself are hers.
Each article starts with a real news event — something she finds genuinely interesting that day. The article URL or PDF goes through a pipeline that reads it, generates a plain-English passage, an explanation, a glossary, and a 10-question SAT-style quiz. She reviews every output before it goes live: checking facts, tightening tone, fixing anything that reads off.
AI does the heavy lifting on drafting and structure. The human in the loop does the part AI still can't do well: deciding what's worth covering and whether each piece is honest enough to publish.
Every article synthesises from real reporting and links back to the original so you can read it yourself. Copyrighted articles are never reproduced verbatim. The source link sits at the top of every piece — click through if you want the full story.
GenAware is in beta. The content library is still being built up before opening fully to the public. Once it launches, the plan is to bring in other high-school students as editors — reading drafts before they publish, fact-checking, and shaping what gets covered. If you're a student who wants in, email ahaanamg@gmail.com.
Python on the back end, Flask + plain HTML/CSS on the front end, SQLite for user data, and Claude (Anthropic's AI) for the writing, quiz generation, and on-page tutoring. Hosted on a small Vultr VPS. The whole stack is intentionally lightweight — fewer moving parts means more time on the content.