Project Info
Inspiration
APIE began with a frustrated teacher: our founder. She watched large language centers charge parents substantial fees without giving them a clear way to see how their children were progressing. Behind the scenes, teachers like her spent hours preparing lessons, grading work by hand, and assembling reports, leaving too little time for what mattered most: guiding students. These were two sides of the same broken system. We built APIE to let AI handle the repetitive work, not to replace teachers, but to give them back their time.
What it does
APIE uses AI agents to support the language-teaching workflow from lesson creation to assessment and progress reporting. For learners: APIE transforms learning materials into interactive lessons, provides on-demand AI speaking practice, and helps learners find a suitable teacher or mentor. For teachers: APIE turns uploaded PDFs and prompts into interactive lessons, evaluates speaking and writing with actionable explanations, records results automatically, and generates progress reports that teachers can share with parents. The system is multilingual and can support different languages, curricula, and proficiency levels.
How we built it
OpenAI models, including GPT-5.6, power the reasoning layer behind APIE’s agents. They interpret uploaded exam PDFs and page images, structure lesson content, evaluate open-ended spoken and written responses, and generate feedback that explains how learners can improve. Azure AI Speech supports pronunciation and speaking assessment, while Imagen 4 generates contextual lesson illustrations on demand. We used Codex throughout the development workflow to implement features, debug agent behavior, refactor the codebase, connect the AI services, and move working prototypes into production-ready product flows. This allowed our small team to test ideas with teachers quickly and improve the product from real classroom feedback.
Challenges we ran into
Our biggest challenge was building an ambitious product with a small team and limited resources. We had to move quickly while balancing product development, user research, classroom testing, and support for hundreds of teachers. That constraint forced us to prioritize ruthlessly. Instead of trying to build every possible feature, we focused on the parts of a teacher’s workflow that consumed the most time: creating lessons, grading open-ended work, and tracking student progress. Codex helped our small team implement, debug, and refine these workflows much faster, allowing us to turn teacher feedback into working product improvements within short iteration cycles.
Accomplishments we're proud of
Built automated speaking and writing assessment with actionable explanations, not just scores Achieved a first-generation lesson acceptance rate above 90% in our pilot, measured by whether teachers saved or shared a lesson without regenerating it Reduced lesson generation time to consistently under one minute Brought more than 400 teachers into the pilot, with another 500 on the waitlist Generated nearly 1,000 worksheets used across classrooms in Vietnam Most importantly, the product supports its own thesis: AI handles the busywork while the teacher remains essential.
What we learned
The biggest opportunity is not simply a smarter AI tutor. It is a lighter workload for every teacher. We also learned that trust is a product feature. Parents engage more when they can see clear evidence of progress, and learners improve more when feedback explains what to do next. A score tells someone where they stand; a useful explanation helps them move forward.
What's next
for APIE APIE, short for AI-Powered Interactive Education, will move its Teacher Tool from a free pilot to a paid launch on August 1, 2026. We are starting in Vietnam, expanding our English offering, deepening our German curriculum, and adding support for more languages before entering international markets. Our goal is to make high-quality, personalized teaching accessible to learners while giving teachers the systems and leverage that previously only large institutions could afford.
This project did not link a GitHub repository.
Analysis
No indexed repository for this project, so there are no commit stats to show.
Technology
- AWSUnchecked
- Google GeminiUnchecked
- OpenAIUnchecked
- VercelUnchecked
No repository was indexed for this project, so these Devpost claims have not been checked against code.
AI coding agents
No repository was indexed, so agent usage could not be checked.
Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.
Codebase size
No repository was indexed, so there is no codebase to measure.
This project did not link a GitHub repository, so there is nothing to diagram.
This project did not link a GitHub repository, so its feature claims have not been checked against code.
Export this project's context (description, README, evidence, key source files) to chat with an AI agent elsewhere.