LessonLens

Inspiration

Teachers often spend hours rewriting one lesson for different learners, then still discover confusion only once class has started. We wanted to move that work earlier: use AI to stress-test a lesson before it reaches students.

What it does

LessonFoundry turns one uploaded lesson into an inclusive classroom package. GPT-5.6 first analyzes the lesson into an editable profile and accessibility score. After the teacher approves it, the app generates standard, ESL, ADHD, gifted, and dyslexia-friendly versions, plus a quiz, homework, and parent summary.

Its signature feature is the Classroom Simulator: GPT-5.6 role-plays five learner profiles, predicts realistic questions, identifies the exact concept causing confusion, and recommends a fix. Teachers can apply a selected fix and refresh only the affected materials.

How we built it

We built LessonFoundry with Next.js 15, React, TypeScript, Tailwind CSS, and the OpenAI Responses API. We use GPT-5.6 for structured lesson analysis, accessibility scoring, differentiated content generation, learner simulations, and targeted risk fixes. Codex helped accelerate the UI architecture, typed API routes, reusable components, and testing.

Challenges we ran into

The hardest problem was avoiding a generic “AI worksheet generator.” We redesigned the workflow so teachers review an analysis before generation, and so simulator feedback is traceable to a specific lesson concept. We also needed the experience to remain useful for a demo when an API key is unavailable, so we added clear sample-lesson states.

Accomplishments that we're proud of

We are proud that LessonFoundry does more than produce content. The Classroom Simulator acts as a pre-teaching stress test: it helps teachers find likely misunderstandings before class, then improve the lesson through an actionable feedback loop.

What we learned

Structured outputs make AI features more usable and trustworthy. Teachers need control, context, and clear recommendations—not a wall of generated text. The strongest use of GPT-5.6 here is reasoning across learner needs and connecting each risk to a concrete improvement.

What's next for LessonFoundry

Next, we want to add document export, richer file parsing for PDFs and slides, saved lesson packages, collaboration for teaching teams, and teacher controls for learner profiles and curriculum standards.

Built With

• codex
• gpt-5.6
• next.js
• openai-responses-api
• react
• typescript