Vibe Coding Built LibraryAid That Sparked Reading

A U.K.-trained primary teacher says her AI-powered reading recommendation system, built through vibe coding, helped unlock new curiosity in students—turning a library catalog into a personalized hunt for books, while keeping data protections and teacher oversi

At 11 p.m., the idea felt urgent. By morning, it had a way of evaporating—until it didn’t.

In May 2025. a U.K.-trained primary school teacher with 11 years of experience in international schools across the Middle East and Southeast Asia decided to stop talking about the gap she kept seeing: librarians could curate books carefully. but too many titles still sat untouched on shelves because there was no systematic way to connect each child to the book most likely to ignite their interest.

So she asked her partner whether she should give up evenings and weekends for a year or two. Her partner’s response was simple: “Just go for it.” By November, her vibe-coded app was live in her classroom.

The premise was direct. Existing solutions were expensive, rigid, and built around proprietary book lists that didn’t match the schools’ real collections. The teacher says the more she looked at what was available. the clearer it became that the problem wasn’t that technology didn’t exist. It was that nothing had been built for teachers like her—working in schools like hers.

She built her own system anyway, using an AI technique she had been reading about with increasing interest: vibe coding.

The months before launch were not smooth. Learning vibe coding meant using AI tools to generate software code by describing what she wanted in plain language. with little to no traditional programming knowledge required. Progress, she says, was painfully slow—moving a day forward, then several days back.

During the summer months, she nearly quit several times. Early architecture decisions haunted her. She was working on a 12-year-old Mac that she hadn’t upgraded, and even installing the right development environment became a full-time job.

One moment still sits at the edge of her memory: several files of code were deleted with no backup. Hours of work were gone. She stared at the screen for a long time.

Then came the hardest part, at least in her telling—book cover images. She wanted the app to display covers for the school’s 10. 000 books using freely available API calls. without scraping. specifically to stay on the right side of copyright laws. Writing the code was exhausting. When it finally worked imperfectly. she built a separate page to manually evaluate every cover—because AI searching for covers that hadn’t loaded correctly took weeks. After that, the page itself failed completely, and she had to start from scratch.

Switching from Copilot to Claude made a significant difference, she says. It was still prone to errors and loops that drove her “absolutely crazy” when she shared the experience with colleagues. But it was more reliable than what she had before.

What surprises her now is the pace. She says what took her days and weeks in late 2025 could later be accomplished in hours, and she describes the improvement in LLMs as “frankly frightening.”

LibraryAid’s process is built around school reality, not re-cataloging. A teacher uploads their school’s library catalog as a CSV file. Then the teacher creates student profiles and runs a short reading assessment to gauge reading level and interests.

From there, the AI analyzes the catalog against each student’s reading level, interests, favorite authors, and curriculum topics, generating a personalized reading list using books already on the shelves.

Student profiles include name, reading age, reading interests, favorite authors, preferred genres, and current class topic. Progress data includes books read, reviews written, points earned, and comprehension quiz scores. The teacher and school librarian are the only people who can see student profiles and progress data—no other students can view them.

When students log in, they see recommendations—typically 50 books ranked by how well they match their profile. Students can mark books as “reading,” “finished,” or “want to read.” When they finish a book, they write a teacher-verified review and answer AI-generated reading comprehension questions.

Correct answers earn genre-specific points. Those points unlock accessories for an animated worm companion—one accessory category per genre across 21 genres—so reading widely is rewarded, not just reading a lot.

Reviews also feed back into the recommendation engine, meaning a “hidden gem” discovered by one child becomes visible to the whole school community over time.

Behind the scenes is a “master books” list built from more than 1. 000 award-winning and highly rated children’s titles across various categories. The teacher says the recommendation engine isn’t just matching reading levels. It is also actively surfacing books most children would never stumble upon independently.

Recommendations can land for different reasons: an award-winning book the child had never heard of; a genre the child hadn’t tried before sitting under a topic they had listed as an interest; or a natural next step—such as a similar author. a continuation of a series. or a book that builds on something the child had already loved and rated.

Data protection was part of the build from the start. She says LibraryAid is COPPA and GDPR compliant, and that student data is stored securely in Google Firebase. No student email addresses are collected. Students log in via a school-issued code and PIN, with no personal email required. Data is never sold or shared with third parties.

The early response, she says, didn’t come only from her own testing.

When she told a colleague what she was trying to do. that colleague’s encouragement gave her more confidence than any tutorial or documentation. The colleague said she genuinely believed the teacher could make it work and that she should not give up. Feedback from other teachers followed, frank and occasionally humbling. One colleague has integrated the app into her class and found it very useful so far.

Her 12-year-old son, she says, became one of her most enthusiastic supporters. He spent considerable time testing the system. told his own school about it. and—she describes it as a distinctly contemporary parenting moment—told his mother he had asked an LLM whether LibraryAid had a high chance of being successful. The LLM responded with an enthusiastic “yes.”.

When the tool went live with her students. she says “something shifted.” Children who had been unenthused about the library before suddenly became excited to explore it. Finding their recommended book turned into a treasure hunt. Students began venturing into new series and authors they would not have chosen independently.

One student stood out in a way she says was hard to ignore: an English learner reading approximately two grade levels below his current placement made 3x the average reading progress of his classmates once he was matched to books that genuinely interested him at the right level. She insists the technology didn’t fix his reading struggles. Instead, it connected him to books worth the effort of reading.

She also read aloud to the class and ended the school year with “Swimming Against the Storm” by Jess Butterworth. which has a strong environmental theme. The impact of reading that book last year. she says. was striking: suddenly the majority of the class was searching the app for adventure stories with a similar feel.

That moment reinforced what she believes deeply about the system: it works best alongside human influence, not instead of it. In her view, the app can surface the right books for students, but it is the teacher or librarian who sparks the interest.

If there is a lesson she carried from building the system into teaching. she describes it as almost the same skill set across very different tasks. Debugging code and diagnosing why a student isn’t understanding a concept require surprisingly similar thinking—being systematic. patient. and hypothesis-driven. Writing algorithms that adapt to different reading patterns made her think more clearly about differentiation. And months spent building something real children would use every day. she says. showed her why so much education technology misses the mark.

In her experience, most edtech is built for administrators, not teachers. It optimizes for reporting and data dashboards instead of the daily reality of 30 children with 30 different relationships with reading. The products that work. she believes. are the ones built by people who have stood in a classroom and felt the gap between what exists and what’s needed.

There is, however, a boundary the teacher is careful to name. Even with enticing books, there is no guarantee students will take action. She remembers a moment with a child this term who showed her her curated list with a lost expression—eyes that were pleading for guidance. The list contained hidden gems and well-known classics with appealing covers. some in her comfort zone and some designed to stretch her thinking. But the only books that interested her were a familiar series she already knew.

The algorithm did its job, the teacher says. What the child needed next was a conversation with a trusted adult.

She argues that no recommendation engine can replace the moment a child says, “I’m not sure about any of these,” and looks to their teacher or librarian for a nudge. The trust a child has for the person standing in front of them, she says, can’t be coded.

Her advice to other educators considering building their own edtech tool is similarly direct: build something that extends what teachers do rather than replaces what they do. Let the technology handle matching. Let children’s learning guides handle the moment.

LibraryAid, she says, turned out to be the most useful thing she has ever built—possibly even eclipsing some of her lessons.

education technology AI reading recommendations vibe coding library catalog personalized reading lists COPPA GDPR classroom innovation international schools