RAG Templates - Enterprise RAG Framework

Production-ready RAG applications with InterSystems IRIS. Zero-configuration APIs, enterprise-grade architecture, and seamless LangChain integration.

🚀 Quick Start

Python - Zero Configuration

from rag_templates import RAG
Works immediately - no setup required
rag = RAG()

rag.add_documents(["Your documents here"])

answer = rag.query("What is machine learning?")

print(answer)

JavaScript - Zero Configuration

import { RAG } from '@rag-templates/core';
const rag = new RAG();

await rag.addDocuments(["Your documents here"]);

const answer = await rag.query("What is machine learning?");

console.log(answer);

ObjectScript Integration

// Direct IRIS integration via Embedded Python
Set bridge = ##class(RAG.PythonBridge).%New()
Set result = bridge.Query("What is machine learning?", "basic")
Write result.answer

📦 Installation

# Python
pip install rag-templates
JavaScript/Node.js
npm install @rag-templates/core

🏗️ Core Architecture

Schema Manager

Centralized schema management with automatic migration support:

• Universal dimension authority for all vector tables
• Automatic schema detection and migration
• Customizable table names and field configurations
• Version tracking and rollback capabilities

IRISVectorStore Interface

LangChain-compatible vector store with enterprise features:

from rag_templates.storage import IRISVectorStore
Drop-in LangChain replacement
vector_store = IRISVectorStore(connection_manager, config_manager)

retriever = vector_store.as_retriever(search_kwargs={"k": 5})

Configuration System

Environment-aware configuration with validation:

from rag_templates.config import ConfigurationManager
config = ConfigurationManager()
Supports RAG_DATABASE__IRIS__HOST env vars
Automatic YAML loading with schema validation

🛠️ Available RAG Techniques

*ColBERT: Includes experimental Pylate integration with pluggable backend support (native/pylate).

🎯 Developer Experience

Three-Tier API Design

Simple API - Zero configuration for prototypes:

rag = RAG()  # Works immediately

Standard API - Production configuration:

rag = ConfigurableRAG({
    'technique': 'colbert',
    'llm_provider': 'openai'
})

Enterprise API - Full control:

config = ConfigManager.from_file('enterprise-config.yaml')
rag = ConfigurableRAG(config)

Key Configuration Options

🔗 MCP Integration

The Multi-Cloud Platform (MCP) integration allows you to easily deploy and manage RAG services as microservices. This design enables flexible deployment across various environments and seamless integration with existing enterprise systems.

Creating MCP Servers

Create MCP servers in minutes:

import { createMCPServer } from '@rag-templates/mcp';
const server = createMCPServer({

name: "knowledge-server",

description: "Company knowledge base",

ragConfig: { technique: 'graphrag' }

});
server.start();

IRIS SQL Tool Integration

The IRIS SQL tool provides direct SQL access and advanced vector search capabilities within your MCP-deployed RAG services. It allows for efficient data manipulation and retrieval directly from InterSystems IRIS databases, leveraging its powerful SQL and vector functionalities. This integration streamlines data management for RAG applications, enabling complex queries and high-performance vector lookups.

For detailed setup and usage, refer to the MCP Integration Guide.

📚 Documentation

✅ Verification

# Quick setup and validation
make setup-env && make install
make validate-iris-rag && make test-unit
Full end-to-end testing with 1000+ documents
make load-1000 && make test-1000
Performance benchmarking
make test-ragas-1000-enhanced

🌟 Key Features

• Zero Configuration: Production-ready defaults, works immediately
• Enterprise Architecture: Schema management, migrations, monitoring
• LangChain Compatible: Drop-in replacement for existing workflows
• Multi-Language: Python, JavaScript, and ObjectScript support
• MCP-First Design: Trivial MCP server creation
• Advanced RAG: 7+ sophisticated retrieval techniques
• Performance Optimized: Built-in caching and optimization

📚 References & Research

RAG Technique Papers & Implementations

Core Technologies

• Vector Databases: InterSystems IRIS Vector Search
• Embeddings: Sentence Transformers, OpenAI Embeddings
• LLM Integration: LangChain, OpenAI API
• Evaluation: RAGAS Framework

🤝 Contributing

We welcome contributions! See our Contributing Guide for details.

📄 License

MIT License - see https://github.com/intersystems-community/iris-vector-rag/blob/master/LICENSE file for details.

Transform your RAG development from complex to enterprise-ready. Start building in minutes, scale to production.