Local Deep Research: AI-Powered, Privacy-Focused Research Assistant for Academia

Introduction

Local Deep Research is an AI-powered research assistant designed for deep, iterative research, achieving approximately 95% accuracy on the SimpleQA benchmark. This robust tool supports both local and cloud Large Language Models (LLMs), enabling comprehensive searches across more than 10 diverse sources, including academic databases like arXiv and PubMed, general web content, and your private documents. With a strong emphasis on privacy and security, Local Deep Research ensures all operations are local and encrypted, making it an ideal solution for researchers, students, and professionals who require accurate, transparent, and secure information retrieval. It systematically breaks down complex questions, verifies information across multiple sources, and generates comprehensive reports with proper citations.

Installation

Getting started with Local Deep Research is straightforward, with several deployment options available. The recommended methods involve Docker for ease of setup.

Option 1: Docker

For a quick setup, you can run Local Deep Research and its dependencies using Docker:

# Step 1: Pull and run SearXNG for optimal search results
docker run -d -p 8080:8080 --name searxng searxng/searxng

# Step 2: Pull and run Local Deep Research
docker run -d -p 5000:5000 --network host \
  --name local-deep-research \
  --volume 'deep-research:/data' \
  -e LDR_DATA_DIR=/data \
  localdeepresearch/local-deep-research

Option 2: Docker Compose (Recommended)

Docker Compose simplifies the deployment by bundling the web app and all its dependencies.

Quick Start (CPU-only base):

curl -O https://raw.githubusercontent.com/LearningCircuit/local-deep-research/main/docker-compose.yml && docker compose up -d

Add NVIDIA GPU Acceleration (Linux only):

After installing the NVIDIA Container Toolkit (refer to the official documentation for detailed steps), use the override file:

curl -O https://raw.githubusercontent.com/LearningCircuit/local-deep-research/main/docker-compose.yml && \
curl -O https://raw.githubusercontent.com/LearningCircuit/local-deep-research/main/docker-compose.gpu.override.yml && \
docker compose -f docker-compose.yml -f docker-compose.gpu.override.yml up -d

Open http://localhost:5000 in your browser after approximately 30 seconds.

Other Options

Local Deep Research also supports installation as a Python package or deployment on Unraid servers. For a complete guide on all installation methods, please refer to the Official Installation Guide.

Examples

Local Deep Research offers flexible interaction through its Python API, HTTP API, and command-line tools.

Python API

Perform quick queries or manage multiple operations with the Python client:

from local_deep_research.api import LDRClient, quick_query

# Option 1: Simplest - one line research
summary = quick_query("username", "password", "What is quantum computing?")
print(summary)

# Option 2: Client for multiple operations
client = LDRClient()
client.login("username", "password")
result = client.quick_research("What are the latest advances in quantum computing?")
print(result["summary"])

HTTP API

Integrate Local Deep Research into other applications using its authenticated HTTP API. For ready-to-use examples, visit the HTTP API Examples directory.

import requests
from bs4 import BeautifulSoup

# Create session and authenticate (simplified for brevity)
session = requests.Session()
# ... authentication logic ...

# Make API request
response = session.post("http://localhost:5000/api/start_research",
                       json={"query": "Your research question"},
                       headers={"X-CSRF-Token": "your_csrf_token"})

Command Line Tools

Manage benchmarks and rate limiting directly from your terminal:

# Run benchmarks from CLI
python -m local_deep_research.benchmarks --dataset simpleqa --examples 50

Why Use Local Deep Research?

Local Deep Research stands out as a powerful and secure AI research assistant due to several key advantages:

• Privacy-Focused: Designed to run entirely locally with tools like Ollama and SearXNG, ensuring your research queries and data never leave your environment.
• Flexible: Supports a wide array of LLMs, search engines, and vector stores, allowing you to tailor the system to your specific needs.
• Comprehensive: Offers multiple research modes, from quick summaries to detailed reports, and can analyze your private documents.
• Transparent: Provides built-in analytics to track costs and performance, maintaining full visibility into the research process.
• Open Source: Released under the MIT license, fostering an active community and continuous development.

Security & Privacy Highlights

The project prioritizes data security with features like:

• Signal-level encryption: SQLCipher with AES-256 protects all user data at rest.
• Per-user isolated databases: Each user benefits from their own encrypted database for complete data isolation.
• Zero-knowledge architecture: No password recovery mechanism ensures true privacy.

Performance & Supported LLMs

Achieving approximately 95% accuracy on the SimpleQA benchmark, Local Deep Research demonstrates strong performance. It supports a variety of LLMs, including:

• Local Models (via Ollama): Llama 3, Mistral, Gemma, DeepSeek, keeping processing local and avoiding API costs.
• Cloud Models: OpenAI (GPT-4, GPT-3.5), Anthropic (Claude 3), Google (Gemini), and over 100 models via OpenRouter.

Extensive Search Sources

Access information from a broad range of sources:

• Free Search Engines: arXiv, PubMed, Semantic Scholar, Wikipedia, SearXNG, GitHub, Wayback Machine.
• Premium Search Engines: Tavily, Google (via SerpAPI), Brave Search.
• Custom Sources: Local documents, LangChain Retrievers, and Meta Search capabilities.

Links

For more information and to get involved with the Local Deep Research project, explore the following resources:

• GitHub Repository: https://github.com/LearningCircuit/local-deep-research
• Official Documentation: https://github.com/LearningCircuit/local-deep-research/wiki/Installation
• Discord Community: https://discord.gg/ttcqQeFcJ3
• Reddit Community: https://www.reddit.com/r/LocalDeepResearch/