TuitionSetu: a tutor's Sunday of WhatsApp updates

Description and use

TuitionSetu automates weekly parent updates for tutors using a local LLM and code validation, cutting preparation time to minutes while keeping data private.

Problem

Tutors spend about two hours each Sunday turning handwritten class notes into personalized WhatsApp updates for parents in multiple languages.

Outcome

TuitionSetu reduces the tutor’s Sunday update time from ~2 hours to a few minutes, ensures data accuracy through code validation, preserves student privacy by running locally, and supports multilingual updates.

Workflow

• 1. Tutor types shorthand notes on phone or laptop immediately after class.
• 2. Local LLM (Gemma via Ollama) parses notes into JSON facts (names, scores, homework, attendance).
• 3. Code validates facts: checks numbers against notes, resolves names, flags unclear items in "Needs you".
• 4. Tutor reviews flagged items, confirms or corrects them, and saves nicknames.
• 5. On report day, tutor opens the weekly review and clicks "Write updates".
• 6. LLM drafts personalized messages in the parent’s language, streaming on screen.
• 7. Code guard checks every number, language, and name usage; flags errors with highlights.
• 8. Tutor edits or dismisses warnings, then clicks "Send on WhatsApp" to open a wa.me link with the message ready.
• 9. Tutor presses Send in WhatsApp; no message is sent automatically by the app.

Evidence

After each batch she jots something like: "Riya ch5 test 14/20 careless in algebra, missed HW twice; Arjun absent"

Source

It takes about two hours.

Source

TuitionSetu does the Sunday part. She types her notes the way she always has, on her phone or laptop. A local model works out who each line is about and every score.

Source

On report day it writes each parent's update in the parent's language. Code checks every number against her notes, and then she reads, approves and taps Send on WhatsApp.

Source

Everything runs on one laptop, with no internet needed.

Source