Create a document and its sources travel with it, every claim bound to where it came from. Query one document or your whole library and get answers cited to the page they stand on, never invented. Integrate the ones people send you, and they answer alongside everything else you already hold.
they don't connect
Your tools collect.
They don't connect.
Most of the work you've ever done is lost. Notes written and never retrieved. Papers read and forgotten. Ideas had in one context and unavailable in another.
AI tools compound the problem. They answer from training data — opaquely, without knowing where the information came from. You can't verify the output. You can't cite it. You can't build on it.
IcebergDocs binds all of it. Every answer traces back to a source. Every source traces back to a document you chose to engage with. Every claim cited to the page it stands on, and the gaps it cannot prove stay visible.
- ▸researchscattered across tabs, folders, and platforms — never retrieved
- ▸pdfsstored, not read. cited to nothing. lost in a folder
- ▸contextrebuilt from scratch on every task, every tool
- ▸ai answersconfident, uncited, and often invented
- ▸documentsshipped dead — no way to ask the reasoning behind them
one document.
From first note to finished deliverable.
One queryable document underneath all of it.
Everything routes through the same cited layer. Context built in query carries into write. Sources found in research become citations in the document. The bibliography falls out complete.
Write it, or drop your sources in.
Author a document, or drop in your readings, notes, and PDFs. Clip a page with the extension, or capture straight from a Claude thread via MCP. Every claim finds its source and binds to it. What comes back is an iceberg: a document with its sources attached, cited and queryable down to the page.
Integrate the ones you receive.
Someone sends you an iceberg. One click folds it into your library, and now it answers questions alongside everything else you hold. Related work lands together: a note from today sits beside a paper from six months ago because they're about the same thing. What you collect compounds instead of scattering.
Answers cited to your sources. Never invented.
Ask a single document or your whole library. Every answer points to the exact source it came from, by reference, down to the page it stands on. If your sources don't have it, it shows you the gap as a research task rather than guessing.
Deep research. Every source logged.
Paste a reference list and IcebergDocs fetches every paper — full PDFs where available, abstracts and links where not. Or run a dossier: an agentic pass that reads your library, finds the gaps, pulls external sources, balances opposing views, and produces a fully cited document. Every source traceable. Nothing fabricated.
29 sources read · 5 external promoted · dialectical balance maintained
Write with every claim bound to its source.
Ask, write, research, and debate share a single persistent context. Switch between stress-testing an argument and drafting a paragraph without losing what you've built. Slash commands let the AI draft, extend, or edit any section — drawing on your sources, your research, and your active thread. As you write, claims bind to their sources and unsupported ones surface as visible gaps.
A queryable document you hand to anyone.
Export any iceberg as a queryable link. Anyone with access can ask the underlying sources a question and get a cited, source-grounded answer — no account required. The reasoning behind every conclusion stays accessible, and the gaps stay visible. Not buried in an appendix.
no card required.
200 sources · 20 queries/day · 10 research runs/day · web + mobile + extension · MCP mount.
Unlimited sources · unlimited query & research · dossier research mode · private spaces · queryable document exports · agent API.