Hey HN,

I built DeepInsight after getting frustrated with how AI research tools treat the research process as a black box.

The Problem - Current tools (ChatGPT, Perplexity, Gemini Deep Research) ask 3-4 clarifying questions, then generate a report you have to manually fix. The human is only involved at the very beginning.

The Solution - Multiple human-in-the-loop checkpoints throughout the research process. You guide the AI at critical decision points, or switch to auto-approve with scheduling when you trust it.

Key features

  - Live collaboration on reports (Google Docs-style)
  - Version control with side-by-side comparison
  - Export: HTML, PDF, DOCX, Markdown, PowerPoint slides
  - Multi-speaker podcast generation → publish to Spotify/Apple
  - Generate FAQs, flashcards, blog posts from reports
  - Chain research from completed reports
  - BYOC (bring your own context/documents)
  - MCP integration, REST APIs, webhooks
  - Team workspaces with sharing controls
  - Prebuilt industry-specific assistants

The Core insight - research quality improves dramatically when humans can intervene at the right moments, not just at the start.

Looking for feedback, especially from researchers, analysts, and content creators.

https://deepinsight.neuralstratlabs.com/