The Information-Theoretic Search Engine for RAG & Agentic Memory

Learn more · Tutorials · Join Discord

Why Moorcheh?

• 32x Compression Ratio over traditional Vector DBs
• 85% Reduced End-to-End Latency over Pinecone vector search + Cohere reranker
• 0$ Storage Cost true serverless architecture scaling to 0 when idle
• Read the full paper

Moorcheh is the universal memory layer for agentic AI, providing fast deterministic semantic search with zero‑ops scalability. Its MIB + ITS stack preserves relevance while reducing storage cost and decreasing latency, providing high‑accuracy semantic search without the overhead of managing clusters, making it ideal for production‑grade RAG, agentic memory, and semantic analytics.

🛠️ Key Capabilities

• Bring any data: Ingest raw text, files, or vectors with a unified API.
• One-shot RAG: Go from ingestion to grounded answers in a single flow.
• Zero-ops scale: Serverless architecture that scales up and down automatically.
• Infrastructure as code: Deploy into your cloud with native IaC templates.
• Agentic memory: Stateful context for assistants and long-running agents.
• Developer-ready: Async support, type hints, and clear error handling.

🚀 Quickstart Guide

Hosted Platform

Use our hosted platform to get up and running fast with managed indexing, zero-ops scaling, and usage-based billing.

Self-Hosted

1. Install the SDK using pip:

pip install moorcheh-sdk

2. Sign up and generate an API key through the Moorcheh platform dashboard.

3. The recommended way is to set the MOORCHEH_API_KEY environment variable:

export MOORCHEH_API_KEY="YOUR_API_KEY_HERE"

Basic Usage

import os
from moorcheh_sdk import MoorchehClient

api_key = os.environ.get("MOORCHEH_API_KEY")

with MoorchehClient(api_key=api_key) as client:
    # Create a namespace
    namespace_name = "my-first-namespace"
    client.namespaces.create(namespace_name=namespace_name, type="text")

    # Upload a document
    docs = [{"id": "doc1", "text": "This is the first document about Moorcheh."}]
    upload_res = client.documents.upload(namespace_name=namespace_name, documents=docs)
    print(f"Upload status: {upload_res.get('status')}")

    # Add a small delay for processing before searching
    import time
    print("Waiting briefly for processing...")
    time.sleep(2)

    # Perform semantic search on the namespace
    search_res = client.similarity_search.query(namespaces=[namespace_name], query="Moorcheh", top_k=1)
    print("Search results:")
    print(search_res)

    # Get a Generative AI Answer
    gen_ai_res = client.answer.generate(namespace=namespace_name, query="What is Moorcheh?")
    print("Generative Answer:")
    print(gen_ai_res)

For more detailed examples covering vector operations, error handling, and logging configuration, please see the examples directory.

API Client Methods

The MoorchehClient and AsyncMoorchehClient classes provide the same method signatures. Below is a list of the available methods.

Methods	Required Parameters	Description
namespaces.create	namespace_name, type, vector_dimension	Create a text or vector namespace.
namespaces.list	N/A	List all available namespaces.
namespaces.delete	namespace_name	Delete a namespace by name.
documents.upload	namespace_name, documents	Upload text documents to a text namespace.
documents.get	namespace_name, ids	Retrieve documents by ID.
documents.fetch_text_data	namespace_name, limit?, next_token?	List text/summary chunks (cursor pagination).
documents.upload_file	namespace_name, file_path	Upload a file for server-side ingestion.
documents.list_files	namespace_name	List raw files in storage for a namespace.
documents.delete	namespace_name, ids	Delete documents by ID.
documents.delete_files	namespace_name, file_names	Delete uploaded files by filename.
vectors.upload	namespace_name, vectors=[{id, vector}]	Upload vectors to a vector namespace.
vectors.delete	namespace_name, ids	Delete vectors by ID.
similarity_search.query	namespaces, query	Run semantic search with text or vector queries.
answer.generate	namespace, query	Generate a grounded answer from a namespace.

For fully detailed method functionality, please see the API Reference.

🔗 Integrations

• LlamaIndex: Use Moorcheh as a vector store inside LlamaIndex pipelines.
• LangChain: Plug Moorcheh into LangChain retrievers and RAG chains.
• n8n: Automate workflows that ingest, search, or answer with Moorcheh.
• MCP: Connect Moorcheh to external tools via Model Context Protocol.

Roadmap (Planned)

Item	Required Parameters	Description
get_eigenvectors	namespace_name, n_eigenvectors	Expose top eigenvectors for semantic structure analysis.
get_graph	namespace_name	Provide a graph view of relationships across data in a namespace.
get_umap_image	namespace_name, n_dimensions	Generate a 2D UMAP projection image for quick visual exploration.

Documentation & Support

Have questions or feedback? We're here to help:

• Docs: https://docs.moorcheh.ai
• Discord: Join our Discord server
• Appointment: Book a Discovery Call
• Email: support@moorcheh.ai

Contributing

Contributions are welcome! Please refer to the contributing guidelines (CONTRIBUTING.md) for details on setting up the development environment, running tests, and submitting pull requests.

License

This project is licensed under the MIT License - See the LICENSE file for details.