I love long-form podcasts, but I’m constantly frustrated by the signal-to-noise ratio. Scrubbing through a 2-hour conversation to find the 5-minute insight is a huge waste of time, and existing search tools just dump you at the start of the episode.
So, I started building Firehose. It’s not another podcast search engine.
Instead, Firehose points you to the key insights, the core ideas, the turning points—and serves them up as short, digestible clips.
The UX is a single feed (think TikTok, but for ideas). You listen or skip, and the recommendation model (in theory) learns what you’re interested in. If a clip hooks you, one tap jumps you to that exact moment in the full episode for context.
My goal is to make those “in-between” moments (waiting for a build, grabbing coffee) actually useful for discovery.
It’s very early and the models are still naive. I’d genuinely appreciate some brutally honest feedback from the HN community.
Does this feel like it solves a real problem for you?
What’s the first thing that seems broken or missing?
Happy to answer any questions. Thank you for your time!
Comments URL: https://news.ycombinator.com/item?id=45868169
Points: 1
# Comments: 1
Source: firehose.exalto.ai
