This repository provides ready-to-run, Docker-friendly application templates for building RAG (Retrieval-Augmented Generation) and AI enterprise search solutions. It targets developers and researchers needing to quickly deploy AI applications that stay synchronized with live data sources like Google Drive, SharePoint, S3, and Kafka, offering significant advantages in data freshness and simplified infrastructure.

How It Works

The core of these applications is the Pathway Live Data framework, a Python library with an embedded Rust engine. This framework handles data synchronization, indexing (vector, hybrid, and full-text search), and API serving in a unified manner. It eliminates the need for separate vector databases, caches, and API frameworks, leveraging in-memory indexing with usearch for vector search and Tantivy for hybrid search. This integrated approach aims for high accuracy and scalability, with templates optimized for simplicity or performance.

Quick Start & Requirements

• Each application template includes a README.md with specific instructions.
• Templates are runnable as Docker containers, exposing an HTTP API.
• Optional Streamlit UIs are available for some templates.
• Supports Linux and macOS.

Highlighted Details

• Live synchronization with various data sources (files, Google Drive, SharePoint, S3, Kafka, PostgreSQL, real-time APIs).
• Built-in in-memory indexing for vector, hybrid, and full-text search.
• Templates include RAG, multimodal RAG (GPT-4o), Unstructured-to-SQL, and private RAG with Ollama.
• Adaptive RAG technique claims up to 4x token cost reduction while maintaining accuracy.

Maintenance & Community

• Active community support via Discord.
• Encourages contributions for documentation, features, and bug fixes.
• Follow on X: @pathway_com

Licensing & Compatibility

• The README does not explicitly state the license.

Limitations & Caveats

• The specific license is not clearly stated in the README, which may impact commercial use or closed-source integration.