
Building a Smart Club Member Portal with AI-Powered Document Search
A smart Club Member Portal that provides secure member access, document-based AI chat, source-backed answers, and real-time document updates for Student Builder Groups.
Introduction
Student Builder Groups need a simple way for members to access club information, workshops, AWS resources, and important announcements. During an event, information can change quickly, and members should not have to search through multiple documents or chat groups to find the latest information.
To solve this problem, we built a Club Member Portal with a login system, member dashboard, document-based chatbot, and document management functionality. The prototype website was created using Lovable, allowing us to quickly design and develop the user interface and application flow.
The project extends the original 70% Club Member Portal and adds the required 30% functionality: an admin document publishing flow, automatic re-indexing, chatbot synchronization, and dashboard updates.
What We Built
Our portal provides the following main features:
1. Member Sign Up and Login
Members can create an account and log in to access the portal. The chat functionality is available only to authenticated members.
2. Forgot Password
Members can request a password reset through their registered email address. This provides a simple recovery flow when a member forgets their password.
3. AI-Powered Club Chatbot
After logging in, members can ask questions about the Student Builder Groups, AWS resources, workshops, hackathon information, and other club documents.
The chatbot is designed to answer questions using the available club documents rather than making unsupported guesses.
4. Source-Based Answers
An important feature of our chatbot is that answers include their source documents. This helps members understand where the information came from and makes the responses more trustworthy.
5. Admin Document Management
Club administrators can add or update documents through the admin side of the portal. Documents can be added by pasting content or uploading a file.
6. Automatic Knowledge Sync
When an administrator publishes a new or updated document, the system re-indexes the content. The chatbot can then use the updated information without requiring the member to refresh the browser.
The target synchronization time for the hackathon is approximately 60 seconds.
7. Updated Member Dashboard
The member dashboard displays the latest document list. When an administrator publishes a new document, the updated document list should appear in the same window without requiring a complete page reload.
How the AI Chat Works
The portal follows a document-based retrieval approach.
First, club documents are added to the knowledge base. When a member asks a question, the system searches the available documents for relevant information. The retrieved content is then used to generate a grounded answer.
The chatbot also returns the relevant source information with the answer.
This approach helps prevent the chatbot from inventing information that is not available in the club documents.
If the required information cannot be found, the chatbot should clearly say that it could not find the information in the club documents and direct the member to the appropriate AWS Student Builder contact.
30% Hackathon Extension
The second phase of the hackathon required us to extend the existing 70% portal rather than create a completely new application.
Our extension focuses on four important areas:
Admin document publishing
Re-indexing after document updates
Chatbot synchronization with the latest documents
Automatic dashboard document updates
The chatbot should use newly published information within approximately 60 seconds, and members should not need to refresh their browser before asking a question about the updated information.
API Integration
The project also supports the required API contract:
POST /ask
The endpoint accepts a question and returns an answer along with its sources.
The response contains:
answer — the grounded answer generated from the documents
sources — the documents used to provide the answer
document — source document name
rank — source ranking
chunk_id and score — optional retrieval information
This allows the evaluator to test the chatbot independently of the website interface.
AWS Deployment Approach
Our prototype can be demonstrated locally, as live AWS deployment is optional for the hackathon.
If deployed to AWS, the system can be mapped to AWS services as follows:
Authentication: Amazon Cognito can manage member sign-up, login, and authentication.
Password Reset: Amazon SES can be used for password-reset emails.
Document Storage: Amazon S3 can store club documents uploaded by administrators.
AI: Amazon Bedrock can provide access to foundation models for generating AI responses.
Knowledge Retrieval: A retrieval-based architecture can be used to search the uploaded club documents and provide relevant information to the AI.
Re-indexing: When an administrator uploads or updates a document, an AWS-based workflow such as S3 + Lambda can trigger processing and update the search index.
Application/API: The chat API can receive member questions, retrieve relevant document content, and return a grounded answer with sources.
This architecture would allow the local prototype to be extended into a scalable cloud-based application.
Demo Flow
Our demonstration follows this flow:
A member opens the Club Member Portal.
The member signs up or logs in.
The member accesses the chatbot.
The member asks a question related to the club documents.
The chatbot returns a grounded answer with its source.
The administrator publishes a new or updated document.
The document is re-indexed.
The member dashboard displays the updated document list.
The member asks a question about the newly published information.
The chatbot answers using the updated document and shows the source.
The required smoke-test questions are demonstrated with pass/fail results.
The /ask API is tested using the required JSON format.
Why This Project Is Useful
The Club Member Portal brings important club information, document search, and AI assistance into one place.
Instead of searching through multiple documents or chat messages, members can simply log in and ask a question. At the same time, administrators can keep the information up to date by publishing new documents.
The source-based chatbot also makes the system safer and more reliable because answers are grounded in the available club documents.
Conclusion
Our Club Member Portal combines a simple member experience with document-based AI assistance. By extending the original 70% portal with admin publishing, automatic re-indexing, chatbot synchronization, and dashboard updates, the system is designed to handle changing event and club information efficiently.
The prototype was developed using Lovable, and the architecture can be extended to AWS using services such as Amazon Cognito, Amazon SES, Amazon S3, AWS Lambda, and Amazon Bedrock.
This project demonstrates how an AI-powered knowledge portal can make information easier to access while keeping responses grounded in trusted club documents.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article