Omphalis - The center of your understanding
by•
Articles, podcasts, videos, documents. Omphalis shows the shape, explains the hard parts, and keeps the moments worth coming back to. Omphalis turns articles, podcasts, videos, and documents into a living Library. It cleans every source, reveals its structure, explains difficult ideas in context, preserves what matters, and connects it across everything you keep. Not another shortcut summarizer—Omphalis helps you understand deeply and brings the right idea back when it matters.
Replies
@stanlymt This is a really refreshing take on knowledge management. I like that the focus is on understanding, not just summarizing. Curious the concept of turning saved content into something you can revisit meaningfully is awesome. Does it also support handwritten notes or just digital content?
@vishnu_kant_07 Thank you! That was exactly the idea behind Omphalis. We spend so much time consuming great content but rarely turn it into lasting understanding.
Today, Omphalis focuses on digital content: articles, podcasts, YouTube, PDFs, newsletters, and your own notes and voice notes. Handwritten notes aren't supported yet, but they're definitely on the roadmap. I'd love to bring them into the same connected understanding over time. Thanks for asking!
@stanlymt That makes sense. I actually like that you're focusing on getting the core experience right first instead of trying to support everything at once. Looking forward to the handwritten notes integration
Turning saved articles, podcasts, and PDFs into structured courses with source citations is pure leverage Congrats @stanlymt🙌. just a Quick que. How do you prevent hallucination or oversimplification when parsing complex technical PDFs?
@priya_kushwaha1 Thanks, really glad the courses angle landed 🙏
Honestly, the thing I'm most opinionated about. Omphalis is grounding-first, not generative. It never rewrites your PDF. The extraction layer cleans and re-lays-out the actual document (columns, tables, the real text), and that real text is what you read. Everything on top- the section map, the marks, the course lessons- points back to a specific place in the source instead of paraphrasing it. So there's no rewritten copy for a model to hallucinate into, and nothing gets simplified away, because the full text is always right there.
A few specifics:
• Structure and chapters are extracted, not invented. Chapters only appear if the document actually declares them and passes a quality check. I'd rather show nothing than fake an outline.
• Notes are tuned for precision over recall (~0.8+ confidence). One wrong note costs more trust than ten missing ones, so it stays quiet when unsure.
• Everything cites its source, and answers are drawn only from what's in your own library, not the open web, so you can always check a claim against the original.
On oversimplification specifically, this is where a piece I'm quietly proud of comes in: memory. You can tell Omphalis what you're working on and what you care about (all visible and erasable in Settings), and it uses that to shape what it surfaces and emphasizes, so a dense paper gets treated toward your intent instead of a one-size-fits-all pass. Courses take it further and let you start at your own level. The principle underneath is "comprehension, not reduction": structure the hard parts so you go deeper, never shrink them so you can skip them. And you stay the judge; everything is keep or dismiss.
Would genuinely love your take if you throw a dense paper at it 🙌
the not another shortcut summarizer line is the whole pitch really, most of these tools optimize for getting through content faster which is the opposite of what you're going for. the part I'm curious about is the connecting across sources over time - if I mark something in a podcast today and read an article that touches the same idea in 3 months, does Omphalis surface that link on its own or do I have to go looking for it?
@omri_ben_shoham1 Nailed it, that line is the whole thesis 🙏 And that scenario is exactly why it exists: it surfaces the link on its own, you don’t go looking. When the article arrives months later, Omphalis matches it against your library by the actual entities and ideas plus semantic overlap, not keywords, and brings that older podcast moment forward. Your mark on it makes it a stronger anchor, so what you chose to care about is what it watches for. It’s deliberately quiet about it (shows up on Today and in Connections, no notification spam), and it gets richer as your library grows. The whole point is catching the connection you’d never have thought to search for.
@Stanly Thomas that answers it well, thanks. the quiet-by-design part is actually what sells it for me, a tool that pings you every time it finds something would get muted within a week. good luck with the launch.
@stanlymt Really interesting approach. Can it combine insights across multiple sources into a single learning path?
@dipanshu_kushwaha5 Absolutely, that's the real thing we are rooting for. Any frontier LLM model can generate a learning path for you, but the real hard challenge is building a cohesive, source-grounded, and precisely sliced lesson plan that matches your intent and objective.