🌟Inspiration

Since I was young, I found my joy in studying, especially when I started to truly understand the material. One of the most memorable experiences I have is studying late at night and getting stuck on something in my study material, with no one to call or text for help. This experience inspired me to create the IKNOW app a study partner that provides help with any questions any time, and without fear of judgment. Whether you need to understand a concept, find examples, get summaries, highlight important parts, or even receive suggestions for potential exam questions, IKNOW is here to support you. The app’s slogan is: "Ask anything, anytime."

🚀What it does

IKNOW serves as an AI-powered study partner that enhances the learning experience through advanced features and intelligent capabilities. Here's what it does:

  • Answers Study-Related Questions: Provides detailed, contextually relevant responses to your questions based on uploaded lecture materials or selected topics.
  • Summarizes Lectures: Extracts key points and concepts from uploaded lecture notes, helping you quickly grasp the most important information.
  • Prepares for Exams: Highlights crucial parts of your lectures that are likely to appear in exams, enabling focused and efficient preparation.
  • Performs Intelligent Web Searches: If an answer isn't found in your study materials, IKNOW performs a filtered and summarized web search to provide accurate external information.
  • Streams Responses in Real-Time: Responses are displayed interactively, allowing for a smooth and engaging user experience.
  • Adapts to Context: Remembers previous queries and adjusts answers based on the ongoing conversation and context for better relevance.
  • Offers a User-Friendly Interface: Makes it easy to upload study materials, interact with the AI, and retrieve the information you need quickly and effectively.

🛠️How we built it

  • Snowflake Cortex Search: Used for efficient retrieval of study materials from a Snowflake database.
  • Snowflake Database: Utilized to store and process study materials, leveraging SQL scripts for schema setup and context filtering.
  • Tavily Search Integration: Added for performing web-based queries when study materials lack answers.
  • Mistral Large-2 Model: Implemented for generating accurate, context-aware responses.
  • TruLens Evaluation: Applied to measure and improve the relevance and groundedness of responses.
  • Streamlit Interface: Developed for user-friendly interaction and real-time streaming of answers.

⚡Challenges we ran into

  • Contextual Relevance: Ensuring that the responses were contextually relevant to the user's query and course material.
  • Streaming Output: Implementing real-time streaming of responses to enhance user experience.
  • Web Search Integration: Filtering and summarizing web search results to provide concise and accurate answers.

🎉Accomplishments that we’re proud of

  • Built IKNOW, an AI-powered study assistant using Snowflake Cortex Search and Mistral Large 2 for context-aware responses.
  • Integrated advanced technologies like Tavily Search for web-based answers and implemented real-time streaming for engaging interactions.
  • Designed user-centric features like lecture summarization, exam prep tips, and chat history.
  • Overcame challenges like ensuring contextual relevance and delivering a scalable, evaluation-driven system using TruLens.
  • Delivered a complete end-to-end solution, from database setup to deployment, making studying more productive and interactive.

📚What we learned

  • Snowflake Cortex: Learned how to leverage Snowflake Cortex for advanced data processing and querying.
  • Evaluation Metrics: Learned how to use TruLens to evaluate the performance of RAG systems based on relevance and groundedness.

🕒What's next for IKNOW

  • Multi-Language Support: Expanding support for multiple languages to serve a global audience.
  • Voice Interaction: Add voice-based assistance for enhanced accessibility, benefiting visually impaired users with voice commands and audio responses.
  • Expand IKNOW to assist professors with creating exams, grading, and providing detailed feedback, streamlining academic assessments.

Built With

  • langchain
  • mistral
  • python
  • snowflake
  • snowflake-cortex
  • snowflake-database
  • sql
  • streamlit
  • tavily
  • trulens
Share this project:

Updates