How to build a Deep Researcher
A 100% open-source, self-hostable Deep Research Stack That Beat OpenAI, Gemini, and Perplexity
A 100% open-source, self-hostable Deep Research Stack That Beat OpenAI, Gemini, and Perplexity
If you need AI to do research for you today, you're probably using ChatGPT Deep Research, Claude, or Perplexity. All three are genuinely capable. All three are also closed-source SaaS running in someone else's cloud.
If you need AI to do research for you today, you're probably using ChatGPT Deep Research, Claude, or Perplexity. All three are genuinely capable. All three are also closed-source SaaS running in someone else's cloud.
Every query you send and every internal document you connect sits on their servers, not yours.
Every query you send and every internal document you connect sits on their servers, not yours.
For most teams, that's been the trade-off: accept it, or don't use AI for serious research.
For most teams, that's been the trade-off: accept it, or don't use AI for serious research.
In this article, you'll see a third option: a fully open-source deep research stack that runs on your own infrastructure.
In this article, you'll see a third option: a fully open-source deep research stack that runs on your own infrastructure.
Three tools, all open source: Onyx for retrieval, CrewAI for orchestration, Voxtral for voice.
Three tools, all open source: Onyx for retrieval, CrewAI for orchestration, Voxtral for voice.
Here's the full system running end-to-end, from voice query to narrated research report:
Here's the full system running end-to-end, from voice query to narrated research report:
0:01 / 0:34
0:01 / 0:34
The rest of this article breaks down how it works and walks you through building the same stack yourself. Before any of that, though, it's worth being clear about why this is worth building at all.
The rest of this article breaks down how it works and walks you through building the same stack yourself. Before any of that, though, it's worth being clear about why this is worth building at all.