Omphalis - Everything you read, understood and yours to keep

by
Articles, PDFs, podcasts, videos and voice notes arrive clean, structured and searchable. Ask questions and check every citation. Select a passage for a real explanation. Write with your sources beside you. Export the whole library as markdown, always.

Add a comment

Replies

Best
Maker
📌

The article you saved last month could help answer today’s question. A passage from a podcast could become the starting point for something you write.

Omphalis helps you return to those ideas, understand them more deeply, and make something of your own.

Since our first launch in July, Omphalis has grown from a reader into a place where your sources, questions and writing stay connected. Here’s what you can do now.

Bring everything you want to understand into one library. Save an article, PDF, podcast, video or your own voice note. Each becomes a calm, readable page, with clutter removed, audio transcribed, speakers identified and the original words preserved. Strata marks the openings, claims and shifts in the piece, helping you find where to spend your attention.

Work through the difficult parts. Select a passage and ask what it means, with an explanation grounded in the text. Ask a question across your library and follow citations back to the exact passages behind the answer.

Explore how your sources fit together. Connections shows where something you saved supports, extends or updates another piece, with an explanation you can examine for yourself. An older article can become useful again when a new question gives you a reason to revisit it.

Develop your own thinking alongside your sources. New this month, Notebooks brings your writing and reference material together. Drag in a passage, type @ to reference a saved item or an earlier answer, and build on it in your own words. When you’re ready, share a read-only link that anyone can open without an account.

Go deeper into something worth learning. Courses turns material already in your library into a sequence of lessons with checkpoints, giving you a path through a subject you want to spend more time with.

There are iPhone and Android apps, and you can export your entire library as Markdown files to take your work wherever you need it.

A few details before you explore: Courses, Connections, Discover, and search by meaning require a paid plan. Courses currently start from saved material; creating a course from an arbitrary topic isn’t available yet.

What’s something in your saved library that you’d love to understand better, connect to other ideas or turn into your own work? We’d love to hear.

What real task does your product handle with GPT-6 Astra?
Straight answer first: Astra built the feature, it does not serve it. Production requests run on GPT-5 class models. The task is course building. Point Omphalis at something already in your library and it returns an ordered set of lessons with checkpoints, written from that source and the reading behind it. It is the most complex subsystem here, and for two weeks Astra and I worked on it. Almost none of that was new features. It was the ways a build fails quietly: sources approved and then lost on the way to being used, a lesson arriving empty, a retry that never ended, a finished course swallowing the next request. Nobody reports those, because the screen looks fine. Finding them meant holding acquisition, narrowing, teachability and delivery in view at once and reasoning about what one stage could do to another. That is what the long context was actually good for, and it is why the courses you can build today finish instead of quietly stalling.