Coming from Gemini Notebook?
Elicana is not a notebook. Some of what you are used to is here, some of it works differently, and some of it we do not do at all. This page says which is which, so you can decide before you pay.
The short version: you point Elicana at the folder where your documents already are. It compiles them once, on your own machine, into a knowledge layer rather than a search index, and then you ask questions across all of them. The cost of a question does not depend on how big the archive is.
What is the same
- You ask questions in plain language and get answers written from your own material, with the passages they came from cited.
- Your source files are never modified. Elicana reads them and keeps its compiled layer separately.
- PDFs, Word documents, spreadsheets, Markdown and plain text all work as sources.
What is different
| In a notebook | In Elicana |
|---|---|
| You upload sources into a notebook, one notebook at a time, and each notebook only sees its own sources. | You point it at a folder, an Obsidian vault, or a Drive or OneDrive folder synced to your machine. The whole archive is one place, and every question can reach all of it. |
| Sources are indexed when you upload them, and each question searches that index for matching passages. Under the new limits, the cost of a question rises with the number of sources in the notebook. | The archive is compiled once into a knowledge layer, not just indexed, and the cost of a question does not grow with the number of documents. |
| Your documents are uploaded to a cloud service and processed there. | It runs on your own machine. With local models through Ollama, your documents and your compiled data never leave it. With a frontier model, your queries go to your provider under your own API key, never through us. You can switch between the two at any time. |
| A search over passages finds the words you typed. An archive written over years uses different words for the same things, so the passage you need may never match. | Compilation reconciles the vocabulary that drifted across years of writing before you ask, so a note from 2019 can answer a question phrased the way you say it now. |
| Per-feature or compute-based limits that reset on a timer. | $20 a month at the early-bird beta rate, flat. We do not pay compute for your questions, so we do not meter them. Up to two machines per subscriber. |
What we do not do
These are the things people ask about most, and the honest answer to each is no, or not yet.
- No audio overviews, no slide decks, no video. Elicana answers questions in text. If the podcast is the feature you use most, we are not a replacement for it.
- Windows and Linux today. macOS is coming soon. There is no web or mobile version; it is a desktop application.
- It is not an offline product. The app checks its license once a day, and that request carries a key and entitlement data only. If you lose your connection it keeps working for fourteen days with a live subscription. Every request the app can make is listed on what leaves your machine.
- Compiling a large archive takes time, and if you choose a frontier model it costs tokens on your own key, once. Local models are free to run and slower. Most people add folders over time rather than all at once.
- Native Google Docs need to be exported first. A synced Drive folder holds Docs as links, not files. Export them as Word or PDF into the folder and they compile like anything else. We are confirming a smoother path and will update this line when it exists.
- It is a beta. Rough edges and fast fixes. Support is direct email at support@elicana.ai, and a person answers it.
If that still sounds like what you were looking for
Register, add a card, and download at elicana.ai/get. The install guide covers first-run setup, and the announcement says what beta means here in more detail.
If it does not, that is a fine outcome too. The point of this page is that you find out here rather than after paying.