An AI teaching assistant for coaching institutes and schools

The problem: teachers spend evenings on paperwork

Teachers in coaching institutes and schools write weekly quizzes, practice worksheets, unit tests, question papers and report-card remarks, and answer parents on WhatsApp. Most of this is repetitive but needs care. Generic AI chatbots can produce questions quickly, but they draw on the whole internet, miss your syllabus and can produce wrong answers with confidence.

How it works

1. Add your material. Teachers upload their own notes, textbooks, past papers and syllabus outlines. Everything is stored per institute.
2. Index it. The system splits and indexes the content so it can find the exact passages relevant to a chapter or topic. This approach is called retrieval-augmented generation (RAG).
3. Ask for what you need. A teacher picks the class, chapter, difficulty, question types and marks, for example "20 mixed questions on photosynthesis for Class 8".
4. Generate from the source. The assistant writes questions and answer keys using only the uploaded material, and shows the source passage for each.
5. Teacher reviews. The teacher edits, swaps or rejects questions, then exports a quiz, worksheet or formatted paper.
6. Remarks and parent updates. From marks and attendance, the assistant drafts personalised report-card remarks and parent messages in English or a regional language. A teacher approves each before it goes out.

Inputs, what the AI does, outputs

Input	What the AI does	Output
Teacher's notes, textbooks, past papers	Indexes content by chapter and topic	A searchable knowledge base per institute
Class, chapter, difficulty, question mix	Writes grounded questions with answers	Quiz, worksheet or question paper draft
Student marks and attendance	Drafts remarks in the teacher's tone	Report-card remarks for review
Test results	Summarises class and student progress	Parent update drafts
Teacher edits	Learns preferred style and wording	Better drafts next time

Integrations

Output can be exported to Word or PDF for printing, or shared digitally. Parent drafts can be sent over WhatsApp Business after approval. Student lists and marks can come from a spreadsheet or your existing school or coaching management software. Built with generative AI on Google Cloud, it can support English and Indian languages, and we test the languages you need before launch.

Guardrails and human review

• Grounded in your material. Questions come from uploaded content, not from the open internet, and show their source.
• Teacher approves everything. Nothing reaches a student or parent without a teacher's review.
• Answer-key checks. Answer keys are cross-checked against the source and flagged if unsure. Teachers should still verify them.
• Student data care. Student information is access-controlled and kept per institute. Because it involves children's data, confirm your obligations under the DPDP Act with your adviser.

Data you need to start

• Digital copies of the textbooks or notes for one subject and class level, with permission to use them.
• Syllabus outlines and a few past papers showing your preferred format.
• A sample of report-card remarks your teachers have written, to capture tone.
• A student list with marks and attendance in a spreadsheet.

A hypothetical walkthrough

A Class 8 science teacher uploads her chapter notes on photosynthesis. She asks for 15 questions: ten multiple choice and five short answer, at medium difficulty. The assistant returns the questions with an answer key and, beside each, the passage it used. She rewrites one ambiguous option, deletes a question outside her syllabus, and exports a printable worksheet. Later she asks for remarks for a student whose marks improved from the previous term. She edits the draft by one sentence and approves. The example is illustrative.

What changes for your team day to day

Teachers move from writing every question to choosing and correcting. A weekly test that took an evening can be drafted in minutes and then checked carefully. Coordinators get consistent formats across teachers, and parents get timely, personal updates without the staff typing each one. Every item still passes through a teacher before use, so quality control stays where it belongs. Teaching, marking and judging a child's progress stay with the teacher.

Rollout

We start with one subject and one class level. Teachers upload their material, and we compare generated questions with what they would have written. After tuning, a small group of teachers uses it for a few weeks and tells us what to change. Then we add report remarks, parent updates and further subjects. Because the product is still in development, exact features and timing will be agreed with each early pilot partner.

Who it fits

Coaching institutes, tuition chains and schools where teachers produce a lot of tests and reports from their own material. See our education industry page, the related admissions calling agent, and Webonika for Schools for the full product direction.