WikiSearch

Inspiration

The inspiration behind this project was to create a user-friendly interface that seamlessly integrates multiple advanced technologies to provide real-time, comprehensive information on any topic from Wikipedia. Leveraging AWS Bedrock for AI-powered summarization, LaunchDarkly for feature flag management, and Streamlit for an intuitive web app, the goal was to streamline the process of retrieving and summarizing information, making it accessible and engaging for users.

What it does

The WikiSearch app allows users to search Wikipedia for any topic and receive summarized results in real-time. Users can type in a query or select from suggested topics. The app fetches relevant Wikipedia articles, summarizes the content using an AI model from AWS Bedrock, and displays the information in an easy-to-read format. The app also uses LaunchDarkly to manage feature flags for dynamically selecting AI models, enabling A/B testing or experimentation seamlessly.

How we built it

Frontend: Built using Streamlit, which provides an interactive and responsive UI.

Added a search bar for user input. Included suggestion buttons for quick queries. Displayed results and summaries in real-time. Backend:

Wikipedia API: Used to fetch articles based on the user's query. AWS Bedrock: Leveraged for AI-powered summarization of the fetched articles. LaunchDarkly: Integrated for feature flag management, allowing dynamic selection of AI models based on feature flags. Session Management: Utilized st.session_state in Streamlit to manage user queries and responses.

Challenges we ran into

Integrating Multiple APIs: Ensuring smooth interaction between the Wikipedia API, AWS Bedrock, and LaunchDarkly was challenging. Real-time Response Streaming: Implementing efficient real-time streaming of AI responses required careful handling of API responses and Streamlit's state management. Session State Management: Maintaining the integrity of session states in Streamlit, especially with dynamic user inputs and feature flags, was complex.

Accomplishments that we're proud of

Successfully integrating advanced technologies (AWS Bedrock, LaunchDarkly, and Streamlit) to create a functional and user-friendly app. Implementing real-time response streaming, providing users with instant and accurate summaries of their queries. Managing feature flags with LaunchDarkly, allowing seamless A/B testing and model experimentation without disrupting the user experience.

What we learned

The power of combining multiple APIs and technologies to enhance the functionality and user experience of a web app. Effective state management in Streamlit is crucial for maintaining a smooth user interface and handling dynamic inputs. Real-time data streaming can significantly improve the responsiveness and interactivity of an application.

What's next for WikiSearch

Enhance Summarization Quality: Continuously improve the AI models and fine-tune the summarization process to provide more accurate and concise summaries. Add More Features: Incorporate additional functionalities like voice search, advanced filtering options, and more interactive elements to enhance user engagement.

Built With

• amazon-web-services
• api
• launchdarkly
• streamlit