The newspaper was a pretty great way to stay informed. Read a good one every day and you could have a reasonable understanding of what was happening in the world. An entire editorial team worked on your behalf: deciding what mattered, giving you enough context to understand it, and fitting it into something you could finish. Keeping you informed was their job.
The internet pulled that bundle apart, and a lot of what followed was wonderful. You can get as much detail about something as you’d like. Specialists can now connect directly with their niche audience. Nothing was gatekept but the incentives shifted as well. Newspapers want to increase readership but they were paid by you the reader. Today’s websites sell their readers’ attention and feed their readers junk food content that keeps them reading but struggles to satiate them.
Readers are now their own editors: finding sources, deciding whom to trust, sorting through what they publish, and figuring out what deserves our time. Each newsletter, podcast, and YouTube subscription adds another source to a disjointed publication we put together ourselves, each component attempting to grab our attention and get paid.
I think AI can give us the best of both these things: by wading through the ocean of information and understanding what matters to each reader, we can build a personalized newspaper built around your work, your interests, what you already know, and what you’re trying to learn. I built FireHose to do just this.
Start with a short brief about what you’re working on, what interests you, and what you’d like to skip. Include the depth you want and the kinds of coverage you find useful. FireHose uses that description to assess incoming material.
For example, someone building an AI product could ask for practical reports of prompt-injection attacks, agent permission failures, and defenses tested in real systems, while skipping product hype. A useful incident writeup should rank above a generic AI security announcement.
Then connect your sources. Websites and blogs come in through RSS or Atom feeds. You can import an existing subscription list as an OPML file, the export format used by many feed readers and podcast apps.
Newsletters arrive by email forwarding. For podcasts and YouTube channels, FireHose uses transcripts to summarize and rank episodes and videos. A recorded interview can appear in the same briefing as an article about the same subject.
You can also follow arXiv categories for new papers and abstracts, or add a particular URL or browser clip. Subscriptions bring in recurring material; saved links let you add something you found elsewhere.
Under the hood, each source becomes text: an article body, newsletter, or transcript. The summary step produces specific takeaways and a fuller account of the piece, with a prompt that asks for names, numbers, and substantive details.
A separate model call compares the summary and takeaways with your brief, returning a relevance score and a short explanation. Separating these steps makes it easier to diagnose whether a poor result came from misunderstanding the source or misjudging its importance to you. The tradeoff is that a detail omitted from the summary may also be missed by the ranking step.
You can read the briefing on the web or set up email delivery. Scan the takeaways, expand a summary, or open the original to read, watch, or listen in full. Where a relevance explanation is available, you can see why an item was selected.
The reader supports highlights and notes, with export for the passages and thoughts you want to keep. Processed material stays available for search, so finding something later doesn’t depend on remembering which feed it came from.
FireHose is currently invite-only. You can request an invitation at firehosesignal.com.