PureRecall Documentation

Welcome to PureRecall Documentation

Sign in to access comprehensive documentation about our RAG search tool

Welcome to the developer documentation for our RAG (Retrieval-Augmented Generation) Search Tool. This documentation provides comprehensive information about the tool's architecture, features, and implementation details.

Please email james@anystudent.ai for access :)

Documentation

Getting Started
Features
Architecture
Implementation
Next Steps
File Overview

PureRecall Documentation

What is RAG Search?

RAG search utilizes semantic vector embeddings to search for relevant information in a document collection. By using an embedding space and an LLM the user can use natural language to search for information with structured and informativeresults. Our tool is highly optimized for the retrieval of transcribed meetings and incorporate a number of "matching" techniques outside of the standard cosine similarity. This is what separates this tool from other generic "only cosine similarity" RAG search tools.

Key Features

  • Vector-based semantic search for improved relevance
  • Automatic document processing and distributed embedding generation
  • Query parsing, keyword extraction and system prompt engineering
  • Hybrid ranking system combining multiple relevance signals like semantic similarity, keyword matching, and temporal relevance

Getting Started

To get started with PureRecall, visit the Getting Started page for setup instructions. Then, explore our Features section to learn about specific capabilities and how to use them effectively.