Building Complete RAG AI Package- AI Curt

#ai #programming #python #rag

Every developer working with Large Language Models eventually hits the same wall. You have the intelligence of a frontier model, but getting that model to talk to your own data securely, efficiently, and without boilerplate fatigue feels like assembling a jigsaw puzzle in the dark.

That exact frustration is what led me to build aicurt, an open-source Python package designed to streamline modern AI integration. And today, I am incredibly excited to share our latest milestone: the release of aicurt v0.1.1, a version built entirely around empowering developers with powerful retrieval foundations and RAG (Retrieval-Augmented Generation) primitives right out of the box.

Why AI CURT?

When aicurt started with v0.1.0, the goal was to build a rock-solid, production-ready scaffold for AI pipelines. Out of the gate, the library shipped with essential utilities that developers usually have to build from scratch:

PII Detection & Redaction Engine: To keep sensitive user data safe before sending prompts.
Chunking & Tokenization Engine: Precise text splitting for context window constraints.
Developer Tooling: Built-in CLI support, automated tests, and streamlined packaging.

Yet, as applications scaled, a critical missing piece became clear: intelligent data retrieval. Developers didn't just need to chunk and sanitize text; they needed an intuitive, high-performance way to search, store, and augment models with that data. That vision drove the development of our newest release.

What’s New in v0.1.1: Laying the Retrieval Foundation

With version 0.1.1, we are taking a massive leap forward in making aicurt a first-class citizen for AI search and RAG architectures. Here is a breakdown of everything new in this release:

Exposed Core RAG & Retrieval APIs at the Root: You no longer need to dig through deep internal directory structures to find what you need. We have exposed the primary classes directly at the package root for clean, Pythonic imports:
- AICurtEmbedder
- AICurtMemoryStore
- AICurtStore
- AICurtRAG
- Essential retrieval helpers
Comprehensive Documentation & Workflows: The README has been completely overhauled. It now features explicit guides and blueprints for implementing keyword, semantic, and hybrid retrieval workflows, giving you immediate patterns to copy, paste, and adapt.
Refined Package Positioning: We have updated our metadata to explicitly position aicurt for robust retrieval infrastructure and advanced AI search capabilities alongside our core privacy and preprocessing features.

Building a Complete RAG Application

With the new public API surface in v0.1.1, setting up a complete, end-to-end Retrieval-Augmented Generation workflow takes only a few straightforward lines of code.

Here is how you can ingest text chunks, manage storage, and query your data contextually:

from aicurt import AICurtRAG, AICurtEmbedder, AICurtStore

# 1. Initialize your core embedder and data store components
embedder = AICurtEmbedder()
store = AICurtStore()

# 2. Spin up the complete RAG pipeline
rag = AICurtRAG(embedder=embedder, store=store)

# 3. Add your custom documents or sanitized text chunks to the knowledge base
documents = [
    "AICurt provides robust PII detection and text sanitization engines for safe AI processing.",
    "Version 0.1.1 introduces powerful retrieval foundations including semantic, keyword, and hybrid search workflows.",
    "Developers can easily orchestrate RAG applications using clean public APIs directly from the package root."
]

# Ingest documents into the RAG pipeline
rag.add_documents(documents)

# 4. Perform high-performance RAG queries across your data
query = "How do I use the new retrieval foundations in aicurt?"
response = rag.query(query)

print("RAG Response:", response)

What’s Next on the Roadmap?

Version 0.1.1 lays down the heavy-duty groundwork needed for reliable, context-aware AI search. But this is just the beginning. Future updates will build upon this retrieval foundation with enhanced caching mechanisms, expanded vector store integrations, and deeper performance optimizations for production environments.

Join the Journey

Whether you are building an internal company search tool, an automated documentation assistant, or securing data pipelines with our PII engines, I would love for you to try out aicurt and see how it fits into your stack.

GitHub Repository: https://github.com/krishnatadi/aicurt
Install via pip:

  pip install aicurt

If you encounter bugs, have feature requests, or want to contribute to the roadmap, please feel free to star the repo or open an issue on GitHub. Let us build better AI search, together!

DEV Community