Case Study - Boosting Classroom Engagement with Generative AI

Overview

Teachers crave interactive lessons but have little time to create them. Our solution solves this by converting a teacher’s natural‑language request (e.g. “Make a counting game for kindergarten”) into a polished, ready‑to‑use activity.

The pipeline:

1. Input – Teacher enters topic and requirements.
2. LLM Reasoning – Claude 3 Sonnet decomposes the request and drafts code.
3. Backend Cleanup – A Rust Actix‑Web service moderates, spell‑checks and sanitises the output.
4. Realtime Frontend – Next.js streams the activity to the browser for live preview and edits.

By embedding activities in SMART Boards and LMSs, Elevate fits seamlessly into existing classrooms.

Key Technical Challenges

• Content Safety & Moderation – Automated filters and human‑in‑the‑loop review.
• Latency – Parallel streaming of LLM tokens and incremental DOM updates.
• Scalability – Provider‑agnostic model layer with fallbacks to OpenAI, Anthropic, Meta and xAI.

What we did

---