VernaSolver: A Retrieval-Augmented Generation

The wide use of AI chatbots has changed the way students look for study help, but these tools answer from the large data they were trained on and not from the exact textbook a student must follow. Because of this, their answers may not match the prescribed book, the page numbers they give may be wrong, and support for regional languages is weak. This paper presents VernaSolver, a study assistant that answers a student’s question only from a chosen textbook and shows the exact page number for every answer. The system uses Retrieval-Augmented Generation (RAG): an uploaded PDF is read page by page, split into small parts, turned into embeddings, and stored in a vector database; at question time the most related parts are retrieved, re-ranked by a second model, and passed to a language model that is instructed to answer only from those parts. When the answer is not present in the book, the system says so instead of guessing. VernaSolver also supports single-book and subject- wide search, replies in English, Hindi and Marathi, and generates quizzes and flashcards from the book. The system was built and tested on real course textbooks, and the results show that it produces grounded, citation-backed answers and reduces the effort of preparing study material.

Authors

Publication Details

Journal
Figshare
Published
2026-11-18
DOI
https://doi.org/10.6084/m9.figshare.35050810
Primary Topic
Artificial Intelligence in Education
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

VernaSolver: A Retrieval-Augmented Generation

Pritam Bairagi
1 links
Figshare
Artificial Intelligence in Education
preprint

VernaSolver: A Retrieval-Augmented Generation

Pritam Bairagi
preprint en

Abstract

The wide use of AI chatbots has changed the way students look for study help, but these tools answer from the large data they were trained on and not from the exact textbook a student must follow. Because of this, their answers may not match the prescribed book, the page numbers they give may be wrong, and support for regional languages is weak. This paper presents VernaSolver, a study assistant that answers a student’s question only from a chosen textbook and shows the exact page number for every answer. The system uses Retrieval-Augmented Generation (RAG): an uploaded PDF is read page by page, split into small parts, turned into embeddings, and stored in a vector database; at question time the most related parts are retrieved, re-ranked by a second model, and passed to a language model that is instructed to answer only from those parts. When the answer is not present in the book, the system says so instead of guessing. VernaSolver also supports single-book and subject- wide search, replies in English, Hindi and Marathi, and generates quizzes and flashcards from the book. The system was built and tested on real course textbooks, and the results show that it produces grounded, citation-backed answers and reduces the effort of preparing study material.

Figshare
Quality education
Artificial Intelligence in Education
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon