Common questions.

Why does comprehension matter?

Because the things you read, watch, and listen to are supposed to change how you think. If you only skimmed the article you loved last week, or let the podcast that shaped your worldview wash over you, or sat through the lecture on something hard without ever really getting it, the time you spent on them paid you back less than it could have.

The world keeps producing content faster than ever, and most tools we use to "save" it (bookmarks, read-later queues, summarizers) help us pretend we engaged with it. Comprehension is the difference between consuming and learning. Between recognizing an idea and being able to use it.

Who is Omphalis for?

For anyone who reads, listens, watches, and wants any of it to stay.

• Writers who need their inputs to feed their thinking
• Teachers and pastors whose teaching depends on what they've taken in
• Researchers whose understanding has to remain retrievable
• Leaders whose decisions depend on real comprehension, not skimmed summaries
• Students and lifelong learners building knowledge that compounds across years

Also the casual reader who simply doesn't want the things that mattered to them to vanish by next month. If you ever caught yourself thinking "I should remember this" about something you read, Omphalis is for you.

Probably, if any of these sound familiar: you bounce off dense articles without really getting them, you skim podcasts and talks that deserved your full attention, or you reach the end of a long piece and realize none of it actually went in.

Omphalis is built for anyone who consumes long content and wants to actually understand it. You don't need to be a researcher or a power user. You need to want to work through what you take in, not just get past it.

Four things, in order.

First, it brings the content in cleanly. Each kind of source (articles, PDFs, podcasts, videos) runs through its own extraction pipeline, so what you get is clean and calmly presented instead of cluttered or fragmented.

Second, it shows you the shape of what you're reading, watching, or listening to: where it begins, where it turns, where it gets dense, where attention should concentrate.

Third, it adds helpful notes where the content gets hard: sparse annotations, only where they earn their place.

Fourth, it keeps the marks you left, with the context they came from. The paragraphs you worked through, the moments you wanted to return to. Over time, those marks connect to other things you've understood, so the library you build actually grows with you.

Summarizers compress content into something shorter. Chat boxes hand you an answer so you can move on. Both make you feel like you understand without doing the work that builds understanding.

Omphalis does the opposite. It doesn't collapse the content. It gives it a structure you can see, and a margin you can mark. You still do the reading. The product does the orienting and remembering.

A summary gives you the gist. Omphalis gives you the shape, the parts that matter, and a way back when you need it.

No. Those are workspaces where you build your own structure from scratch — empty pages, your own links, your own filing system. Omphalis starts with content that already has structure (an article, a podcast, a PDF) and exposes it. You don't have to organize anything. You mark what matters and the system holds the rest.

People who love note-graph tools can still use them. Omphalis sits earlier in the pipeline — it helps you understand and keep what you read, before you decide what to do with it.

Those tools save things and remind you of highlights. They are good at saving and at surfacing snippets — but they do not help you understand a long piece in the first place, and the highlights they show you are out of context.

Omphalis does the work earlier. It maps the shape of the piece before you read it, helps you read it well, and keeps the moments you choose to mark with the context they came from. When something comes back to you later, it comes back connected.

You can. We use language models too. The difference is what we use them for.

Generative tools optimize for fluent answers. They tend to replace your reading with their output, which feels efficient but builds nothing — you walk away with an answer you did not earn, and the understanding does not stay with you.

Omphalis uses language models to do the unglamorous work: read the shape, find the dense moments, suggest annotations where you actually need them. The system's suggestions are suggestions, not verdicts. Your reading, your marks, your decisions stay yours.

You do not have to choose. Omphalis is available inside ChatGPT, so your library and its structure are there in the conversation.

Not really, and the difference is worth being direct about. NotebookLM is excellent at what it does: you give it a set of sources and it generates things from them, like summaries, audio and video overviews, study guides, and reports. It is a source-to-output workspace.

Omphalis works earlier, and closer to the act of understanding. It cleans and lays out long content so you can actually follow it, shows you the shape, adds sparse notes where the hard parts are, and keeps the exact moments you mark. Over time those marks become return points and connections across your own content.

We do offer summaries. Their job is to help you navigate, prioritize, and return, never to reduce a piece so you can skip it. The whole product is built to stay true to the content, not to replace reading it.

The short version: NotebookLM turns sources into outputs. Omphalis helps you stay with the source and keep what matters.

The first layer is clean content. What this takes depends on the format, so each kind of source has its own extraction pipeline: articles are stripped of clutter and boilerplate, PDFs have their layout and columns untangled, podcasts and videos are transcribed and cleaned into readable text. The result is clean content and a calmer, clearer presentation, whatever you brought in. Comprehension starts before the first note, so getting the source right comes first, and the whole pipeline rests on it.

The second layer is structure. Omphalis turns a long piece into a map of typed moments: the opening, the turn, the dense section, the practical implication, the final question. You see the shape before you commit your attention.

The third layer is annotation. Where the content gets dense, unfamiliar, or makes a claim that needs support, Omphalis adds a short helpful note in the margin. Notes are sparse on purpose. One wrong note damages trust more than ten missing ones.

The fourth layer is connection. The articles, podcasts, videos, and PDFs you mark over time start to find each other. Something you read in February becomes available when you open something related in November. This layer compounds. It is most useful after you have used Omphalis for a while.

A mark is the moment you decide something is worth keeping. It is the act of attention — a paragraph, a sentence, a timestamp, an idea — that you want to be able to return to.

Marks are not highlights. Highlights tend to pile up unread; you make them quickly and lose them quickly. Marks are smaller in number and load-bearing — they are the things you actually want the future-you to find. Omphalis builds your library around them.

Discover is a shelf inside your Library that suggests reading related to what you already keep, with a line on each card explaining why it came up. Every item is rights cleared, and you can keep one into your Library in a single action.

Discover is included on Pro and Premium. On Free the shelf is visible as a preview.

A course is a ladder of objectives in the order you have to learn them, each one taught from a real source rather than generated from a prompt. Rights are settled before any material is extracted, non-commercial and unverifiable sources are refused, and nothing is rewritten: a lesson is a real passage from a real source, and quiz answers are quoted from it. Every course shows the sources it was built from, with each licence.

When the available sources cannot support a subject, the builder says so instead of building anyway.

Reading courses is included on every plan. Building your own is a Premium feature.

Inside Learn Omphalis, the built-in guide, you can ask how something works and get an answer grounded in the guide itself, with links to the section it came from. It covers the product.

Questions about your own saved content are answered through search and Connections.

Spaces let you keep separate threads of your understanding. Each thing you save lives in one Space, and you switch between them across Today, Inbox, Library, and Connections, so your work reading stays clear of your personal reading, and one topic does not blur into another.

Spaces are free on every plan.

Currents is a section on Today that shows where your library meets what is happening now. When a new development relates to something you have kept, Currents surfaces it and names the relationship: what it challenges, what it updates, what it applies to, or what it continues, always linked back to the item in your library.

Currents is part of Connections, so it is included on Pro and Premium. It draws on what you have saved, so it grows sharper as your library does.

The pile of good intentions is the problem every read-later tool leaves you with. Omphalis handles it quietly. Anything you saved and never opened gradually moves to a Set aside shelf, so your library stays the pieces you actually engaged with, instead of everything you ever clicked.

Nothing is deleted, and nothing you touched is moved. The moment you mark, note, star, or even open something, it stays put. Whatever is set aside has its own shelf you can browse any time, and one tap brings any of it back.

This is on for everyone, free included.

Articles, podcasts, YouTube videos, PDFs, EPUBs, Word docs, newsletters, your own notes, and voice notes. You can paste a URL, upload a file, or add an RSS feed.

Voice notes and meetings can also enter the same pipeline — your own thinking is treated like any other piece of content, with the same maps, marks, and connections.

Paywalled content you don't have access to. Live broadcasts. DRM-protected files. Encrypted PDFs without a password. Anything Omphalis cannot legitimately read on your behalf.

Yes. Omphalis uses language models to read the shape of a piece, find the dense moments, and suggest notes where understanding usually breaks. We're careful about what we ask them to do, and careful about what we don't.

Most tools that say "powered by AI" replace your thinking with a model's output. A summary, an answer, a verdict. Omphalis does the opposite. The system shows you the shape of what you're reading and suggests notes where they earn their place. Every decision is still yours. Your marks are yours. Your return points are yours. The system's suggestions are suggestions, not verdicts.

If Omphalis ever feels like it's thinking for you instead of with you, that's a bug. Tell us.

Your saved content, marks, and notes live in a database that is yours alone. Each user has their own isolated storage. We do not pool your content with other users' content.

No. Your content is used to serve your experience — to generate the maps, annotations, and connections you see. It is not used to train models for us or anyone else.

Yes. You can export everything you have saved, marked, or noted at any time. There is no lock-in.

You get your data out, in a clean format you can take to another tool. We are committed to making sure the things you marked do not disappear because we did.

Omphalis keeps a small, structured memory so it can pick better moments for you: the topics you keep returning to, and on higher plans, how you read and what you are working toward.

You can see every entry it holds in Settings and remove any of them, and removal takes effect immediately. The memory holds labels and weights, never the text of what you saved.

Free uses one memory layer, Pro two, and Premium all four.

Yes. The Free tier covers around 8 articles, or a couple of podcasts, a month with full structural comprehension, 100 subscribed sources, 2 GB of storage, and an hour of Everyday listening. It also includes Spaces, the first memory layer, and reading any course in the catalog. Enough to decide whether Omphalis earns a place in how you read. Your library and your own notes are never capped on any tier.

Yes, and more than a trial. While founding spots remain, sign up and you join as a founding member: 90 days of full Premium (full Connections, full listening hours, full storage), plus early access to new features and a direct line to the team. Once founding spots fill, new accounts start on a 30-day Premium trial instead. After your Premium window, the account moves to Free, with everything you saved kept, unless you choose a paid plan.

Roughly 60 articles or 16 podcasts or 8 PDFs a month, 1,000 subscribed sources, 25 GB of storage, 4 hours of Everyday listening, and Connections-light: entity links plus semantic search across your library. Pro also adds Discover, Currents, and a second memory layer. See the Pricing page for the full feature list.

Premium adds full Connections (claims, edges, themes), fuller Currents, all four memory layers, the ability to build your own course from openly licensed sources, unlimited sources, 50 GB of storage, and HD + Lifelike listening hours, for people who want the deepest layer working across everything they save. A Scholar tier for students and academics is architecture-ready but not launched. We'll ship it when there's enough interest to sustain it.

Because every save runs the full comprehension pipeline (Strata, annotations, semantic indexing) on real language models, not a cheap summarizer. "Unlimited" offers in this space usually mean one of three things: shallow models doing surface-level work, your data being sold or used to train someone else's product, or venture funding subsidizing the run until it isn't.

We'd rather charge for what the work costs and tell you exactly how much depth you've got. The cap is what makes the quality of one item honest. 40 units a month on Free is small, but every one of those 40 saves gets the same processing that a Premium save does. Less volume, same depth. Never the other way around.

Three things, all of which cost us real money.

Comprehension. Every save runs language-model calls: Strata (the structural map), annotations (the dotted underlines), and on paid tiers, Connections (cross-content links). Real model calls on real providers. We pass the cost through at a tier-flat rate rather than per-call surprises.

Listening. When you tap Listen on an article, podcast, or note, we synthesize audio. TTS is billed to us per second of generated speech, so it's metered on your side too. That's why Free gets 60 minutes and Premium gets 8.5 hours of higher-quality voices.

Storage. Uploads (PDFs, images, voice notes) and the audio we generate for you. Capped per tier because R2 storage isn't free for us, but generous enough that a typical year of active use doesn't bump the ceiling.

Things we will never charge for: reading what you've already saved, your own writing, your own notes, your highlights, your tags, exporting your data, or access to your account. Those stay free at every tier, including after a subscription lapses.

What we won't do: sell your data, train models on your reading without consent, run ads against your library, or surveil your engagement to optimize anyone else's product. You are not the product here.

If models get cheaper, faster, or better (and they have been, steadily), we widen the caps before we widen the margin. The unit count is the lever: your $9/month or