Knowledge Bridge: Conversational SharePoint Access Using Amazon Bedrock
Project phases
CUCCIO is the Canadian University Council of Chief Information Officers, a group that advances the sharing of knowledge and resources across IT leaders. To help make its information easier to navigate, CUCCIO and the University of British Columbia’s (UBC) Cloud Innovation Centre (CIC) developed a solution that connects a conversational interface to CUCCIO’s Sharepoint – helping CUCCIO and its members find, and discover, information more easily.
The prototype is powered by large language models (LLMs) and connects a conversational interface with a SharePoint repository. Leveraging the organization’s Microsoft Entra IDs, users can ask questions in natural language and receive relevant responses, based on appropriate access permissions. The summarized answers, sourced references, and role-based access controls mean that users can find information more quickly, and CUCCIO’s shared knowledge base can be used more effectively.
Approach
To create the prototype, the UBC CIC incorporated role-based access management by leveraging existing Entra groups. These groups manage permissions and control access to information based on a person’s affiliation or role, ensuring users access only the information they are authorized to view, according to their role.
The prototype leverages large language models to interpret user queries, retrieve relevant information, and generate summarized responses with sourced references through a natural language, conversational interface. The solution runs on Amazon Web Services (AWS) and uses key services including Amazon Bedrock (see Technical Details).
The team also built administrative features, including an analytics dashboard to monitor platform usage and engagement trends. User feedback can also be collected and used to identify opportunities for ongoing platform improvements.
For more technical details, please visit the GitHub repository.
Screenshots of UI
This section shows the platform interface for users and system administrators.
User View
Administrative View
Supporting Artifacts
Click below to view the prototype’s technical details, including its architecture, or go directly to the GitHub repository.
Architecture Diagram
Technical Details
The solution runs on a serverless Amazon Web Services (AWS) architecture, enabling administrators to manage SharePoint Data Sources and users to get AI-powered answers – all without managing any servers or infrastructure.
AWS Amplify hosts a React application that serves as the user interface and automatically deploys it whenever developers push code changes. Amazon Cognito handles authentication, letting users sign in with their existing Microsoft organizational accounts rather than creating new ones. All backend communication goes through Amazon API Gateway, with AWS WAF protecting against malicious traffic and keeping the service reliable.
User data, such as conversations, settings, and system prompts, are stored in Amazon RDS for PostgreSQL. AWS Lambda functions handle all backend logic and connect to the database through Amazon RDS Proxy, which manages connections efficiently so the system stays responsive under load.
Administrators can load content from Microsoft SharePoint directly from the dashboard. An AWS Glue job handles the heavy lifting: it reads the documents, processes them, and stores them as searchable embeddings using Amazon Bedrock – so the content library stays current without any manual data preparation. AWS EventBridge and Amazon SNS notify administrators the moment an import finishes or fails, and Amazon CloudWatch lets them monitor live progress logs.
When a user asks a question, the system finds the most relevant content from the knowledge base – but only content that user is permitted to see, based on their Microsoft Entra group memberships. This means responses are automatically scoped to each user’s access level without any extra configuration. That context is then passed to an Anthropic Model on Amazon Bedrock, which streams a grounded response back in real time, keeping answers accurate rather than relying on the model’s general knowledge alone.
Acknowledgements
This project was developed in collaboration with CUCCIO.
Student Team: Development by Hrishi Logani and Aniket Nemade. Project assistance by Carrie Schulz.
Image courtesy of CUCCIO.
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.
