The challenge
Implementing AI for various data connectors across multiple websites. The complexity arose from ensuring that the AI could handle and integrate data from diverse platforms like Discord, Twitter, Google Docs, and Notion. Implementing prompt engineering was particularly intricate, as it required the AI to generate accurate and relevant responses to user queries, tailored to the specific type of content. Additionally, analyzing each website to accommodate the unique types of questions users might ask added another layer of complexity. Storing all this data in a database for efficient retrieval further compounded the challenge, as it required meticulous organization and optimization to handle the diverse and extensive datasets effectively.
The goal
The primary goal of this project were to create a robust tool that seamlessly connects and queries data across multiple platforms and document types. The project aimed to simplify the process of importing raw data from various sources, such as PaperArxiv, Knowledge based web, Google Docs, and Notion, and to enable users to ask precise, content-specific questions. By implementing advanced prompt engineering, the project sought to deliver accurate and relevant responses to user queries. Another key objective was to ensure that all generated outputs were efficiently stored in a database for easy retrieval and display.
The result
The project was successfully implemented, achieving its goals of efficiently fetching user queries and obtaining data from multiple sources. It effectively integrates information from diverse platforms ensuring that users receive accurate and relevant responses tailored to their specific needs. The tool's ability to handle a wide range of data connectors and utilize advanced prompt engineering has proven successful in providing users with precise information and insights, affirming the tool's effectiveness in meeting user expectations and requirements.
What was built
Features:
- Query Handling for Various Document Types: Obtain and analyze queries from standard document files.
- Image File Queries: Retrieve and process queries related to image files.
- PowerPoint Slides Queries: Analyze and respond to questions about PowerPoint slides.
- Audio File Queries: Process and provide insights based on audio files.
- Book Queries: Handle queries from complete books, including chapter-based inquiries.
- String File Queries: Analyze and answer questions related to string value paragraphs.
- Google Drive Links: Retrieve and analyze data from files linked via Google Drive.
- Amazon S3 Bucket: Access and query data stored in Amazon S3 buckets.
- Unstructured File: Retrieve query data stored in unstructured files.
- And Many More…
Technology:
- Backend: Flask Framework, Python, Llama & GPT Index Libraries
- Frontend: Angular, Typescript
- Prompt engineering API: OpenAI Data Connectors
- Hosting: AWS (Amazon Web Services)
- Database: SQL for storing query data, and performing CRUD operations on essential data
