You're about to bet a year of your life on a market thesis. The analyst reports are eighteen months old, the blog posts are content marketing, and your network gives you three enthusiastic opinions from three people who'd love to be right. What you actually want is unglamorous: what do the people operating in this market say about it when they talk for an hour?

That conversation exists — hundreds of times over. Operators, investors and burned founders dissect every market on podcasts and conference stages with a candor that never makes it into writing. The consensus (and the dissent, which matters more) is sitting in those transcripts. Nobody has time to watch two hundred hours of it. You don't have to.

deep_query is built for exactly this shape of question. Give it your market thesis as a research prompt and it runs the full pipeline — search, select, transcribe, retrieve, generate — across the spoken record, then answers with inline citations pinned to the second of each source video.

The answer you get back isn't an average of takes. Ask for the disagreement explicitly and it maps the terrain: who says the market is saturated and why, who says the incumbents are vulnerable and where, which risks come up in every conversation and which only worry one loud person. For a decision this size, the shape of expert disagreement is worth more than any single opinion — and you get it in minutes, verifiable at every claim.