Open Education AI Study Companion

Project phases

Published: December 16, 2025

Last Updated: 3 weeks ago.

Recognizing an opportunity to enhance open educational resources in British Columbia’s post-secondary education system, BCcampus partnered with the UBC Cloud Innovation Centre (CIC) to develop an AI-powered study companion. The platform enables instructors to generate supplementary teaching materials based on open educational resources (OER) which students can interact with in an engaging chat interface. By advancing the use of open educational resources, BCcampus aims to help remove barriers to learning and promote innovative pedagogical approaches.

The AI-powered study companion prototype enables instructors to generate practice materials such as quizzes and flashcards to support instruction with OER and guides students through conversational prompts to enhance their learning. The use of open web content and the release of the solution as open source reflect BCcampus’s commitment to fostering innovation and inclusivity in teaching and learning. This platform is reusable and adaptable for other educational contexts—such as K-12 classrooms, vocational training, corporate learning environments, and global OER initiatives—creating a foundation for scalable, AI-enhanced learning experiences.

Approach

The web application is built on AWS Cloud Infrastructure and uses generative AI to support intelligent interaction with OER textbooks. By leveraging Retrieval Augmented Generation (RAG), multimodal embeddings, and conversational AI, the platform provides students with personalized learning assistance. This application’s User Interface (UI) is informed by Universal Design Learning (UDL) Principles with the goal of supporting learner agency by promoting purposeful and reflective engagement, resourceful exploration, and action-oriented learning.

Within the student interface, the AI-powered Study Companion provides real-time responses with citations included for guidance. Students can adjust audio settings that allow narration or autoplay with options to alter voice rate, pitch, and volume. Instructors can use the platform to create material types in 3 formats: Multiple Choice Questions (MCQs), Short Answers, and Flashcards. Short Answers support AI-assisted grading and feedback, while MCQs offer feedback based on the correctness of student answers. Instructors can use the Material Editor to review and revise practice materials. Edited materials can be exported as H5P packages, which initiates a server-side packaging workflow and generates a downloadable zip file compatible with LMS platforms such as Canvas or Moodle. Administrators have oversight of shared content and platform settings, ensuring consistent alignment with open educational goals.

Click here to go directly to the project GitHub repository.

Screenshots of UI

This section outlines the core stages of the user journey, as represented through screenshots of the user interface.

STUDENT VIEW

Home page — _{Students can browse the textbook library and select the subject they want to explore.}

Chat — _{This opens a page that allows students to chat, access FAQs, review practice materials, and use guided prompts to support their learning.}

_{If students wish to share their chat conversation with their instructor or a classmate, they can create a public URL for their conversation using the Share button. Viewers who open the link can preview their chat.}

_{Students can access a list of commonly asked questions along with their answers through FAQ. Students can also report questions if they see something administrators should be aware of, such as inappropriate language and irrelevant questions.}

_{Students can access study materials such as Multiple Choice Questions (MCQs), Flashcards and Short Answer questions through the Practice Material tab. The generated materials stay within the session and can be exported. MCQs provide instant feedback and short answers support AI-assisted grading.}

_{If students are unsure of where to start, they can browse previously used prompts that have been shared by other users through Shared Prompts. Inline prompt cards can be inserted as prompts into current chat sessions.}

INSTRUCTOR VIEW

_{In the Material Editor, Instructors can review and refine MCQ, Short Answer, and Flashcard sets. They can adjust questions, reorder items, add explanations, and export the final set as an H5P package for LMS use or as a PDF for either student or instructor formats.}

ADMIN VIEW

_{Administrators are able to manage multiple settings and monitor usage in a dashboard which provides a platform-wide overview, including total users, questions asked, and active textbooks.}

_{Administrators can view all textbooks with their status, usage, and controls for enabling, re-ingesting, or deleting entries within Textbook Management. They can quickly search, update, and manage textbooks from this table.}

_{Administrators can manage the system prompt, set token limits, and update operational settings through AI Settings.}

_{Administrators have access to the usage and activity metrics such as chat counts, prompt usage, and practice material generation activity.}

_{Administrators can view, dismiss or delete reported FAQs and shared prompts through FAQs and Prompts.}

Supporting Artifacts

Click below to see technical details of the solution, including the detailed Architecture. Or click here to go directly to the project GitHub repository.

Architecture Diagram

A detailed explanation of the diagram can be found on the project GitHub repository.

Technical Details

The platform runs on a secure, scalable AWS stack: The React frontend (Vite) deploys via Amplify and communicates with backend services through Amazon API Gateway using REST and WebSocket protocols, while AWS Cognito provides admin authentication, and a lightweight public JWT flow issues short-lived tokens for anonymous users. Backend functionality is implemented as Lambda functions (REST handlers, WebSocket connect/default, and Dockerized text-generation workers), with API Gateway routing and custom authorizers ensuring proper access control.

Amazon RDS (Postgres) stores chat sessions, interactions, and analytics behind an RDS Proxy. The database uses pgvector for embedding storage and efficient vector similarity queries. Retrieval-Augmented Generation (RAG) pipelines query the Postgres vectorstore to collect textbook context. The system then invokes Amazon Bedrock LLMs for generation and citation-aware responses. The WebSocket layer streams live LLM responses to clients. Dockerized Lambdas handle long-running and asynchronous text-generation jobs using queued invocation patterns.

Administrators upload textbook CSVs via pre-signed S3 URLs. This triggers ETL and ingestion jobs using AWS Glue and SQS that crawl uploaded content, chunk text into manageable segments and generate embeddings. The system persists embeddings and metadata back to RDS, keeping the retrieval index current. The platform implements security through Secrets Manager and SSM Parameter Store for credentials and model configuration, and IAM roles follow least-privilege principles. CloudWatch provides observability through logs and alarms that surface errors and performance metrics.

Acknowledgements

This project was funded by the William and Flora Hewlett Foundation and created in collaboration with BCcampus.

Image by Zen Chung.

Student Team: Development by Harsh Amin, Aniket Nemade and Ayush Srihari. Project assistance by Anya Ameen.

About the University of British Columbia Cloud Innovation Centre (UBC CIC)

The UBC CIC is a public-private collaboration between UBC and Amazon Web Services (AWS). A CIC identifies digital transformation challenges, the problems or opportunities that matter to the community, and provides subject matter expertise and CIC leadership.

Using Amazon’s innovation methodology, dedicated UBC and AWS CIC staff work with students, staff and faculty, as well as community, government or not-for-profit organizations to define challenges, to engage with subject matter experts, to identify a solution, and to build a Proof of Concept (PoC). Through co-op and work-integrated learning, students also have an opportunity to learn new skills which they will later be able to apply in the workforce.

Projects