Omphalis
01 / 33

Omphalis

Long content, mapped.

A comprehension companion for the articles, podcasts, videos, PDFs, and notes you already care about.

Scroll

The short version

Omphalis gives long content a map, a margin, and a way back.

Map

See the structure

A summary gives the gist. A map helps you follow the content — where it turns, where it thickens, where to re-enter.

Margin

Mark what matters

A margin lets you respond. Understand the dense parts, and declare — in your own hand — what mattered.

Way back

Return & connect

The useful parts stay reachable. Over time they connect to everything else you've read, heard, and thought.

The problem

People take in meaningful content, then lose the useful parts.

They read. They listen. They watch. They save. They highlight.

Then it disappears — into memory, tabs, bookmarks, scattered notes, and half-remembered fragments. The problem isn't only overload. It's that nothing helps the content you chose become clearer, markable, and returnable.

The lived pain

Long content breaks in predictable ways.

People don't say "I need a comprehension system." They say this:

"I listened to that whole episode, but I lost the argument."
A two-hour podcast becomes a half-remembered feeling.
"I know I read something about this before."
A saved article becomes another unread link.
"My saved things just become a pile."
A dense PDF becomes a wall of text.
"I highlighted it, but I never used it."
A highlight becomes a dead fragment, cut from its source.
"I need to understand this, not just finish it."
A summary gives the gist, but not the structure.

The category gap

Current tools solve pieces, not the whole loop.

Each category is good at one move. None of them keep you with the content, then bring you back to it.

What Omphalis is

A comprehension companion for long-form content.

It doesn't own the content. You bring articles, podcasts, videos, PDFs, newsletters, notes, and highlights from where they already live. Omphalis makes them easier to enter, understand, mark, and return to.

  1. 01See the shape.The landscape of a long thing, before you enter it.
  2. 02Understand the hard parts.Help exactly where the content gets dense.
  3. 03Mark what matters.Declare meaning in your own hand.
  4. 04Find your way back.Marked moments become return points.
  5. 05Connect meaning over time.Across everything you take in.

The product model

Four layers of understanding.

Omphalis applies four layers of intelligence to the content you bring. The first two work on item #1; the last two compound over your whole library. Layer 1 is the foundation; Layer 4 is the moat — and your marks are the user act that threads through all of them.

Layer 1 · on arrival

Clean

Any source enters cleanly — clutter stripped, layout untangled, made readable.

Before any AI output

Layer 2 · the shape

Structure

Every piece gets a navigable shape — typed moments, density — before you enter it.

Works on item #1

Layer 3 · while reading

Annotations

Sparse, high-precision help, only where understanding usually breaks.

Works on item #1

Layer 4 · over time

Connections & return

What you marked becomes reconnectable across your whole library.

Compounds at 50+ items

Layer 1 · arrives clean

Comprehension starts before any AI output.

Most products treat ingestion as plumbing. Omphalis treats it as the first layer — any source comes in cleanly, its clutter stripped and its layout untangled, before it is ever structured or annotated.

Articlesarrive full of clutterbecome a clean, readable page
PDFshave broken layoutsbecome followable columns
Transcriptsare noisy and timelessget diarized and split at pauses
Podcastsare a flat, linear hourgain a navigable shape
Notesdrift away from sourcesstay tied to the moment that prompted them

Layer 2 · see the shape

A two-hour podcast becomes followable.

Instead of a flat two-hour timeline, you see the structure: typed moments, dense passages, the exact second worth marking. This is orientation, not summary — scrub it.

Layer 3 · understand the hard parts

Helpful notes, only where it's dense.

Three to five high-precision enrichments per piece — Source, Define, Expand. Never opinions, never filler. A bad annotation is worse than none, so Omphalis stays quiet unless it helps. Hover a dotted phrase.

Essay · the center of things · 6 min

Finding the stone

The omphalos at Delphi was a carved stone the Greeks treated as the navel of the world — the center of every distance, the fixed point from which meaning radiated outward.

To read well is to find that stone inside a text: the place an argument actually turns. Most tools hand you the whole quarry. Omphalis points to the dense moments where a careful note earns its place — and stays quiet about the rest.

Source · Define · Expand — hover any dotted phrase

The mark · your act, not a layer

The mark is the irreducible unit of Omphalis.

A mark isn't a fifth layer — it's the user act that threads through all four. A model can summarize the source, or guess what matters.

But your mark says what mattered to you. That is the difference between model-generated importance and user-declared meaning — and it is the seed every connection in Layer 4 grows from.

Most tools hand you the whole quarry. Omphalis hands you the one stone worth carrying — and lets you say, in your own hand, exactly what it meant to you.

Click a soft-underlined phrase, then choose a meaning

A mark can mean

Important
Confusing
Use later
I disagree
Question
Changed my view

Each mark stores its source, its location or timestamp, your intent, an optional note — and becomes a return point and the seed of future connections.

Layer 4 · return & connect

Marked moments become return points.

Not generic document similarity. The strongest connections grow from what you marked, questioned, saved, and returned to — across every kind of content. Hover a node.

The felt experience

"I can see this now."

Not "AI summarized it for me." The first time someone opens a long thing and sees its shape, the reaction isn't admiration for the model. It's relief.

  • ·Ten major moments
  • ·Where the argument begins
  • ·Where the dense section starts
  • ·Which claim may need context
  • ·The exact timestamp worth marking

Positioning · what it is not

Not a better NotebookLM.

NotebookLM is powerful, and we respect it. It turns a source set into generated outputs. Omphalis does the opposite — it helps you stay with the source.

NotebookLM

Turns sources into outputs.

  • Source-grounded chat
  • Audio & video overviews
  • Reports & study guides
  • Mind maps, flashcards, quizzes
Omphalis

Helps you stay with the source.

  • Clean presentation
  • Structure maps
  • Sparse annotations
  • Your marks & return points
  • Connections from marked moments

Positioning · what it is not

Not another chat box.

Most AI tools begin with a prompt. Omphalis begins with the content itself — never framed as "ChatGPT for your documents."

A prompt-first tool

Upload → prompt → answer.

  • Starts with a blank box
  • You do the asking
  • The content is just context
Omphalis

Content → map → marks → return.

  • Its shape, up front
  • Its dense moments
  • Your marks
  • The way back

Positioning · what it is not

Not "Pocket with AI."

Read-later tools solve capture. Omphalis solves comprehension and return — a new experience for long content, not a smarter queue.

Read-later

Saves the link.

  • Capture
  • A growing queue
  • Eventually, a pile
Omphalis

Helps you enter the content.

  • Clean reading
  • Structure & dense moments
  • Annotations & marks
  • Return points & connections

Positioning · what it is not

Not a domain tool.

Zotero, Obsidian, Readwise, Logos — strong where they specialize. Omphalis doesn't compete there. It works across the content you actually encounter, and across time.

Domain tools

Go deep inside one field.

  • Reference management
  • Personal knowledge bases
  • Highlight sync
  • Powerful, but bounded
Omphalis

Works across sources & moments.

  • Articles, podcasts, videos
  • PDFs & newsletters
  • Notes, highlights, voice
  • Continuity across all of them

Positioning · what it is not

Not ambient memory.

Ambient tools watch everything and ask "what did you see?" Omphalis preserves the meaning you chose, and asks "what mattered?"

Ambient memory

What did you see?

  • Watches screen activity
  • Transcribes meetings
  • Captures everything by default
Omphalis

What mattered?

  • Explicit curation
  • You decide what enters
  • It keeps what you chose

Who Omphalis is for

For people who learn from long-form content.

The behavior matters more than the title — a sustained learning practice, real habits: saving, highlighting, subscribing, and quietly complaining about forgetting what they read. The older docs call them the Committed Learner.

Writers Builders Teachers Leaders Researchers Students Pastors Serious podcast listeners

The business model

We charge for intelligence, never for your own thinking.

Every account opens on a 30-day Premium trial — no card. After that, three plans metered by comprehension units, the work of turning a source into something understood. Reading what you've saved, your own notes, and exporting your data are free at every tier.

Free

Available now
$0

 

Start with the things you actually chose.

40 units / mo · ~8 articles
30 premium reads · 30 min transcription
  • Clean reading and structure on every item
  • 100 subscribed sources · 2 GB storage
  • Library kept forever. Voice notes never metered.
  • Extension, mobile share sheet, full export

Pro

Available now
$9/ month

$90 / year · save 17%

Every article, podcast, and video you take in, understood.

300 units / mo · ~60 articles
150 premium reads · 80 PDF pages · 10 hrs transcription
  • Definitions, sources, expansions at the dense moments
  • Mark what matters and add your own notes
  • Connections light: entity links and semantic search
  • 1,000 sources · 25 GB · add a pack anytime

Premium

Available now
$19/ month

$190 / year · save 17%

The whole library working together.

600 units / mo · ~120 articles
400 premium reads · 200 PDF pages · 25 hrs transcription
  • Cross-content connections across everything you saved
  • Engaged-item scoring across your recent top 100
  • Everyday + HD + Lifelike listening hours
  • Unlimited sources · 50 GB storage

We charge for what costs us money — synthesis, storage, intelligence. Never for access to your own thinking.

What the allowance buys

Premium extraction, not cheap scraping.

Comprehension is only as good as what comes in. Each format runs its own high-fidelity pipeline before a single layer of intelligence touches it — the extraction allowance on each plan is how much of that work it includes.

Cheap scraping

  • Grabs raw HTML — ads, nav, and boilerplate ride along
  • PDFs collapse into jumbled, out-of-order text
  • Auto-captions: no speakers, no punctuation, drifting timing
  • Paywalled and JS-rendered pages come back empty
  • Garbage in — every layer above inherits the mess

Omphalis premium extraction

  • Article body isolated — clutter and boilerplate stripped
  • PDF columns, tables, and reading order untangled
  • Audio & video diarized, punctuated, split at real pauses
  • Hard sources rendered and read the way you'd see them
  • Clean source in — Strata, annotations, and connections stay honest

Unit economics & lean infrastructure

Built as a focused, durable business.

  • Low fixed infrastructure
  • Edge-native architecture
  • Usage-based processing
  • Priced by comprehension units
  • Credits extend runway
  • A small team, by design

A durable SMB, not a unicorn-by-default

500–1k
paid users · year one
10k
paid users · year three
25–50k
paid users · year five

Core infrastructure: Cloudflare Workers, D1, R2, Vectorize, Workers AI, Durable Objects, Queues.

The moat

The moat is accumulated, user-declared meaning.

Features are a time-delay advantage. The durable moat lives inside the account — and a competitor can copy the product but not the months of a particular person's attention.

Short term

Clean extraction quality. Structure quality. Annotation trust. Real, but copyable.

Long term

Marks, notes, return history, accepted and rejected connections, voice reflections — cross-source understanding that only grows with time.

Why the moat compounds

The replacement cost is time, not money.

A competitor can copy every feature in a quarter. It cannot import what it never held:

Your marks Your notes Voice reflections Return history Connection decisions An evolving line of thought

Rebuilding those takes months of re-engagement — which is exactly why marks and return points are first-class from day one.

The model strategy

Models are replaceable workers.

Not "send content to a frontier model and show the answer." Frontier or fine-tuned models generate candidates; Omphalis owns the comprehension standard and decides what surfaces. The edge isn't a bigger model — it's a better pipeline.

The intelligence stack owns

Structure schemas Annotation gates Abstention logic OCEval Gold Set The mark graph Source intent Return logic

The go-to-market motion

Don't market the app. Market what it produces.

The best asset is an Omphalis map of a real article, podcast, video, or PDF. For creators: let Omphalis map your long work for your audience. Show value first; invite people in second.

  • 01Public content mapsArtifact-led posts, not feature ads.
  • 02Creator collaborationsMap a real episode or essay for their readers.
  • 03Early-access testsOne real content item per tester.
  • 04Serious long-form usersThe people who already feel the pain.

Launch validation

The next proof is small and concrete.

Don't validate with compliments. Validate with behavior — did they understand the map, did they mark anything, would they return to it?

30
qualified early-access leads
15
real content tests
10
users create at least one mark
5
say they'd use it again
3
say they would pay
5
ask for the extension or source automation

Roadmap

Prove the loop. Build the companion. Build the moat.

Phase 1 · now

Prove the loop

  • Clean content
  • Structure maps
  • Sparse annotations
  • Marks
  • Return points
Phase 2 · next

Build the companion

  • Browser extension
  • Timestamp marks
  • Source intent
  • Inbox Ready
  • Marked view
Phase 3 · later

Build the moat

  • Connections from marks
  • Spaces
  • Return surface
  • Fine-tuned quality gates

What is already built

Omphalis is not starting from zero.

The work now isn't merely building software. It's sharpening the launch loop around comprehension — on top of a real foundation.

Web app iOS app Desktop app Browser extension Content engine Audio engine Cloudflare infrastructure RSS & article pipelines PDF processing Highlights & annotations Semantic search

Why this team can build it

Built by a product-minded engineering leader.

Stanly Thomas

Stanly Thomas

Founder & Engineer · Voxiven

Twenty-plus years of engineering leadership across web, mobile, desktop, browser extensions, infrastructure, and AI workflows — currently at Paylocity, after Microsoft, Oracle, and Schneider Electric. The credibility is execution depth, not founder theatrics.

Paylocity · now Microsoft Oracle Schneider Electric

The ask

Help Omphalis prove the comprehension loop.

  • Creators — let Omphalis map one piece of your long work for your audience.
  • Users — try it with one real article, podcast, video, or PDF.
  • Design partners & early users
  • Infrastructure, model & evaluation credits
  • Content collaborations
  • Patient angel capital, aligned with a durable prosumer outcome

Not everything you see.
Everything you chose.

Appendix · in one breath

Ten seconds

Long content, mapped. Omphalis helps you see the structure, understand the hard parts, and mark what matters.

Thirty seconds

Omphalis makes long content easier to understand. It gives articles, podcasts, videos, and documents a clean page, a clear map, helpful notes where things get dense, and a way to mark the moments that matter. Over time, those marks become return points and connections across what you read, listen to, watch, and think.

Appendix · reference

The details, kept close.

NotebookLM, exactly

Source set → outputs

NotebookLM: source set becomes generated outputs. Omphalis: content itself becomes a clean map, annotations, marks, and return points. Different lane — content engagement, not study artifacts.

The mark model

What a mark stores

Source · location or timestamp · context · user intent · optional note · return point · future connections. A mark can mean: important, confusing, use later, I disagree, question, changed my view.

Inbox · Library · Marks

Custody vs depth

Inbox: what arrived. Library: what was chosen. Marks: what mattered. Source intent sets how much attention future arrivals deserve. Processing depth is separate from custody.

Legal & trust posture

Not a shadow library

Omphalis is not a downloader. Content can stay where it lives. It preserves links, metadata, timestamps, marks, notes, lawful transcripts, and return points.

Discipline

What Omphalis will not do

No social feed. No infinite scroll. No engagement-farming. No leading with AI branding. It won't replace domain tools, download restricted media, compete on generic summaries, or become a creator audio studio.

Why not ChatGPT or Claude?

Before and after the question

They're excellent when you have a question. Omphalis helps with everything around it: entering content, following structure, noticing hard parts, marking meaning, returning later, and connecting marked moments over time.

Omphalis

Voxiven LLC · 2026 · omphalis.ai