Spandan is a real-time classroom polling platform built to make teaching sessions more interactive and give educators a clear window into how their students are actually doing — not just who raises their hand.
Spandan is a web-based EdTech platform designed for classrooms and live presentations. It lets educators run live polls, collect student responses, generate questions from lecture audio, and track engagement — all within a single tool.
The platform is built around two core ideas: reducing the effort it takes for a teacher to ask a question mid-session, and giving students a reason to stay engaged rather than zone out.
Most classrooms still rely on one-way communication — a teacher speaks, students listen, and engagement is assumed rather than measured. The real problems are:
- Teachers don't know if students are keeping up. There's no easy, non-disruptive way to check understanding mid-lecture without stopping the flow entirely.
- Creating good questions takes time. Writing MCQs, True/False or discussion prompts before every session adds to an already heavy workload.
- Participation is uneven. In large groups, the same few students respond while the rest disengage quietly.
- Existing tools are either too simple or too complex. Basic polling apps lack depth; LMS platforms are too heavy for a quick in-class check-in.
Spandan tackles this by making it dead simple to fire off a question mid-session, generate questions from lecture content, and see in real-time who's engaged and who's lost.
Live Polling — Teachers create a room, share a code, and students join instantly. Questions go live and responses come back in real-time — no page refreshes, no delays, no friction.
Audio to Text - Spandan transcribes the real time audio, and based on the segment time it sends the transcript to the LLM for quality question generation.
Questions Generation — Spandan uses the live transcription that is coming from the lecture and based on it, the system suggests relevant questions. Teachers can review, tweak, and push them live in seconds. Saves a lot of prep time.
Question Approval Workflow — Auto-generated questions can go through a quick review before students see them. A small but important step that keeps quality in check and in this way we are enforcing the human in the loop.
Leaderboards & Scoring — Students earn points for correct answers and the leaderboard updates as responses come in. Adds a bit of healthy competition. Teachers can keep it anonymous if the class needs that.
Analytics & Results — Once a session ends, teachers get a breakdown — how the class responded, who scored what, which questions tripped people up. Enough to be useful for getting insights in order to improve the lectures for next time.
Teachers & Educators — The whole product is designed around their workflow. Create a room, add or generate questions, go live, review results. That loop is kept as short as possible on purpose.
Students — Join with a code, answer questions as they appear, watch the leaderboard. The experience is meant to stay out of their way and let them focus on participating.
Live session flow:
Teacher creates a room → shares room code → students join
↓
Teacher start the mic button in the app and start teaching
↓
Transcription happens in real time
↓
Question generation from the Transcription
↓
Teacher sends a question live
↓
Students respond in real-time
↓
Results appear on the teacher's dashboard
↓
Leaderboard updates → session ends → data saved
Audio-to-question flow:
Teacher start lecture and audio get captured
↓
Spandan transcribes based on the segment time
↓
Questions are suggested from the segments transcript
↓
Teacher reviews and approves
↓
Question get pushed in room to student to answer it
This is where the interesting engineering work lives. If you're picking this up, here's what needs thinking through:
- How do you keep question delivery and response collection truly instantaneous? Any noticeable lag breaks the classroom experience. Think about what communication model fits best here.
- How do you transcribe audio reliably without depending on a paid third-party API? The transcription needs to work offline or self-hosted. What model would you use, and where would it run?
- How do you handle state across a live session? When a teacher pushes a question, every student in that room needs to see it immediately. How do you keep everyone in sync, especially with unreliable connections?
- How do you score and rank students fairly in real-time? Responses come in at different times. Does speed matter? Should it? How do you calculate and update the leaderboard without it feeling laggy?
- How do you separate what a teacher sees from what a student sees? Both roles use the same platform but need completely different views and permissions. What's the cleanest way to enforce that?
- How do you make the audio transcription pipeline feel fast? Audio capture, segmenting, transcribing, and suggesting questions is a multi-step process. How do you keep the teacher in the loop without making them wait at every step?
Displaying Spandan_Product.md.