Prototype

Deep research

A multi-source research engine that searches, scrapes, transcribes, and synthesises content into structured digests.

Why we built this

Research is fragmented. You search Google, watch YouTube, check Twitter, read articles, and try to synthesise it all yourself. We wanted one interface that hit every source in parallel, transcribed videos, tracked signals, and saved everything structured.

How we built it

React frontend, FastAPI backend. Four parallel sources: Perplexity AI for editorial content with citations, ScrapingBee for Google and news, YouTube API with auto-transcription, Twitter API for social signals. Time-range filtering (breaking to all-time). Insights sidebar tracks volume and sentiment. Select any text to save to your knowledge base. Digests save as timestamped JSON.

What's next

We use this for client research and competitive intelligence. Fast, comprehensive, structured output. Not packaged as a product yet but considering it if we can simplify the interface for non-researchers.