deep research / agentic RAG on private data
DeepSearcher is an open-source deep research agent that combines LLMs and vector databases (e.g., Milvus) to search, evaluate, and reason over private data, producing answers and reports.
Same problem, same approach. Swapping one for another is a config change, not a rewrite.
Choose for a more mature and widely adopted deep research agent with broader integrations, if you do not need Milvus-specific optimization.
An autonomous agent that conducts deep research on any data using any LLM providers
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Choose for a JavaScript/TypeScript implementation with SSE and MCP support, if you are building in the Node.js ecosystem.
Use any LLMs (Large Language Models) for Deep Research. Support SSE API and MCP server.
Choose if you are already on the OpenAI Agents SDK stack and want a lightweight iterative-research reference, but it is less flexible for private data.
An implementation of iterative deep research using the OpenAI Agents SDK
Choose for a multi-agent LangGraph/LangChain design with credibility scoring, if you want explicit orchestration and citation scoring.
Multi-agent autonomous research system using LangGraph and LangChain. Generates citation-backed reports with credibility scoring and web search
Choose for a progressive-search and cross-evaluation framework, if you prefer iterative hypothesis refinement over vector-DB retrieval.
A deep research framework based on progressive search and cross-evaluation.
Choose for a state-of-the-art web-focused deep research agent from Tongyi, if your use case is public web research rather than private data.
Tongyi Deep Research, the Leading Open-source Deep Research Agent
Choose for fully local web research with no external APIs, if you need to run everything offline and do not need vector-DB private data features.
Fully local web research and report writing assistant
Choose for optimized web research and prediction tasks with strong benchmark scores, instead of private-data-centric workflows.
MiroThinker is a deep research agent optimized for complex research and prediction tasks. Our latest models, MiroThinker-1.7, achieves 74.0 and 75.3 on the BrowseComp and BrowseComp Zh, respectively.
Choose for a report-generation pipeline that produces structured, citable articles, rather than interactive Q&A on private data.
An LLM-powered knowledge curation system that researches a topic and generates a full-length report with citations.
Choose for a minimal, widely-forked TypeScript iterative research bot, if you want a simple hackable JS/TS codebase.
An AI-powered research assistant that performs iterative, deep research on any topic by combining search engines, web scraping, and large language models. The goal of this repo is to provide the simplest implementation of a deep research agent - e.g. an agent that can refine its research direction overtime and deep dive into a topic.
Choose for a local-first research platform with verifiable multi-step workflows and cloud/local model switching.
Cerno is a local-first research platform that leverages agentic AI to break down complex queries into verifiable, multi-step workflows. Switch seamlessly between cloud LLMs and self-hosted models, track every reasoning step, and optimize cost and tokens—all while keeping your data on your machine.
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
Choose as a simpler, older Python reimplementation of deep research; good for learning, but it lacks private-data and vector-DB features.
Deeper Seeker is an simpler OSS version of OpenAI's latest Deep Research feature in ChatGPT.It is an agentic research tool to reason , create multi step tasks , synthesize data from multiple online resources and create neat reports
Does the same job, but ships as an app. Useful to a person, not swappable into a codebase.
Self-hostable personal AI assistant with deep research as a feature; choose if you want an end-user app with UI and API.
Your AI second brain. Self-hostable. Get answers from the web or your docs. Build custom agents, schedule automations, do deep research. Turn any online or local LLM into your personal, autonomous AI (gpt, claude, gemini, llama, qwen, mistral). Get started - free.
These projects were analysed and named deep-searcher among their alternatives. The relationship is not symmetric — how deep-searcher rates them is a separate judgement, made when deep-searcher is analysed in its own right.
calls deep-searcher “Use when you want a deep research tool focused on reasoning over private data with integrated vector search.”
An LLM-powered knowledge curation system that researches a topic and generates a full-length report with citations.
calls deep-searcher “Choose when you need open-source deep research and reasoning over private data.”
Chat with your documents on your local device using GPT models. No data leaves your device and 100% private.
calls deep-searcher “deep research over private data with RAG; choose it when your source is internal documents, not the web.”
Tongyi Deep Research, the Leading Open-source Deep Research Agent
calls deep-searcher “focused on searching private data rather than the open web”
An AI-powered research assistant that performs iterative, deep research on any topic by combining search engines, web scraping, and large language models. The goal of this repo is to provide the simplest implementation of a deep research agent - e.g. an agent that can refine its research direction overtime and deep dive into a topic.
calls deep-searcher “Deep research over private data instead of the web; use when your sources are internal documents.”
Fully local web research and report writing assistant
calls deep-searcher “Use if your deep research must reason over private/local documents rather than the public web.”
MiroThinker is a deep research agent optimized for complex research and prediction tasks. Our latest models, MiroThinker-1.7, achieves 74.0 and 75.3 on the BrowseComp and BrowseComp Zh, respectively.
calls deep-searcher “When your deep research must reason over private/corporate data using vector search.”
Your Personal AI super intelligence. A brain that builds a local-first memory of your life, a fantastic orchestrator of agent fleets and workflows, and a deep researcher.
calls deep-searcher “focused on deep research over private data, choose for research-only workloads”
Fully Local Manus AI. No APIs, No $200 monthly bills. Enjoy an autonomous agent that thinks, browses the web, and code for the sole cost of electricity.
calls deep-searcher “narrower: deep research over private data producing reports, not papers.”
Fully autonomous & self-evolving research from idea to paper. Chat an Idea. Get a Paper. 🦞
calls deep-searcher “Deep research on private data only; choose when your corpus is local documents and you don't need web search.”
~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.
calls deep-searcher “It is a deep-research application over private data, not a RAG library.”
RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs
calls deep-searcher “an end-user deep research tool for private data”
Ready-to-run cloud templates for RAG, AI pipelines, and enterprise search with live data. 🐳Docker-friendly.⚡Always in sync with Sharepoint, Google Drive, S3, Kafka, PostgreSQL, real-time data APIs, and more.
calls deep-searcher “End-user deep research app; choose if you want a standalone research tool for private data.”
Complete API layer for private AI applications on local models: RAG, skills, tools, MCP, text-to-sql, and more. Works with any OpenAI-compatible inference server.
calls deep-searcher “Choose when you need a deep research tool to reason and search over private data, as an application.”
[EMNLP2025] LightRAG: Simple and Fast Retrieval-Augmented Generation
calls deep-searcher “Choose it when you want an out-of-the-box deep-research searcher over private data rather than a graph-RAG library.”
A modular graph-based Retrieval-Augmented Generation (RAG) system
calls deep-searcher “Choose for a deep-research search tool over private data; it's an application, not a library for building RAG.”
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG