01
What it is
A continuously updated library of what the web knows — web pages, research papers, news and documents — read in full and stored so any question can be matched against it in milliseconds.
The Web Index
A live indexof the web, readfrom everywhere.
Most AI products rent someone else's search results. askFinz reads the web itself — and not from one place. The browser extension, the desktop app, devices running askFinz OS, dedicated indexer machines and partners feeding their sites directly all read into the same index, so it stays current without waiting for anything to be rebuilt.
Go deeper: the appliance · extension · search · partners
Connecting…
—
Pages indexed
—
Machines reading
—
Cores pooled
—
Web read
index · liveweb/
web/
├── read by/
│ ├── extension as you browse
│ ├── desktop app its own browser
│ ├── askfinz os on the device
│ ├── indexer nodes dedicated
│ ├── crawler public + partner
│ └── partners direct, priority
├── handles/
│ ├── pages cleaned
│ ├── documents page by page
│ ├── app pages built first
│ └── hard cases last resort
└── becomes/
├── meaning not keywords
├── citable back to source
└── instant no rebuild waitread onceanswer always
Start here
in plain termsThink of it as askFinz's library — a working copy of the useful web that it keeps, reads and maintains itself.
Pages, research papers, reports and documents are read all the way through, understood, and filed so they can be found again in an instant. Nothing sits on a shelf going out of date: a page read a minute ago is already available to answer with.
So when you ask askFinz a question, it isn't recalling something it once memorised. It looks the answer up in this library and shows you the page it came from.
The short version
- It's ours. Not rented from another search company.
- It's fed from everywhere. Every askFinz surface reads into it.
- It's current. Pages are searchable the moment they're read.
- It's deep. Documents get read, not just linked to.
- It's honest. Every answer points back to its source.
02
What's in it
Millions of pages, tens of thousands of scholarly works, and thousands of documents that were opened and read rather than merely listed. It grows every second the fleet is running.
03
How askFinz uses it
Ask a question in Chat, Research, Search or News and the answer is looked up here first, then written with a link back to the page it came from — so you can check it rather than trust it.
Sources indexed
Where it comes from
06 ways in · one indexEverything askFinz runs on is also reading.
A conventional search index has one way in: a crawler on a server somewhere, working through a list. askFinz has six. Every surface the product runs on — in your browser, on your desktop, on a device, on a dedicated machine — reads into the same place.
That is why it stays current without waiting for a rebuild, and why it reaches pages a single crawler never would.
What isn't in it
the public web onlyYour browsing is not the index.
Reading “from everywhere” means the public web is reached from many directions — not that anything private is collected along the way. This index is built from pages that are already published for anyone to read.
How we handle your dataYour history isn't collected
Visiting a site can tell askFinz that the site is worth reading. What then gets read is the pages that site publishes openly — not the page you happened to be on.
Nothing behind a login
If a page needs an account, a password or a session to see, it isn't read. That includes your own askFinz workspaces.
The whole picture
05 places to goHow a page becomes an answer
seen → read → understood → answerableFour steps, and no waiting in between.
A traditional index rebuilds on a schedule, so what it knows is always a little out of date. Here the last step happens the moment the first one does — a page seen now is answerable now.
01
Seen
A page is reached — by the crawler working through the web, or by the extension as somebody reads it.
02
Read properly
Menus and clutter come off. Documents are read all the way through. Pages that only assemble in a browser get built first.
03
Understood
What the page means is worked out and stored — not just the words on it, so a question can match an idea rather than a keyword.
04
Answerable
It becomes searchable immediately, and any answer drawn from it cites the page it came from.
Live · search performance
from the running systemIndexed once, answered instantly.
The whole index is searched on every question. These figures aren't from a slide — they're read from the engine while you look at them.
Connecting…
—
Pages indexed
live, always growing
—
Cold query
fresh, never seen before
—
Cached query
popular / repeat
—
Cache hit rate
served without re-search
—
GPU-embedded
with CPU fallback
query → vector on the GPU pool · payload-indexed SafeSearch filter · deterministic result cache
Live · what's in there
depth, not page countMost of what matters isn't a web page.
Research papers, reports, filings and slide decks hold some of the most useful knowledge there is. Most search engines will happily link to a PDF without ever having read it.
Connecting…
—
Scholarly works
open-access research, indexed in full
—
Documents read
not just linked to
—
Characters from documents
read page by page, not skimmed
—
Document formats
converted to readable text
Live · the dedicated machines
one of the six ways inThe machines that do nothing but read.
Alongside the surfaces that read as part of normal use, a set of computers is given over to the job entirely — some lent by people with capacity to spare, some installed with partners. These are their figures, live.
Add yoursConnecting…
—
Memory pooled
across the machines
—
Machines online
dedicated to reading
—
Processor cores
spare capacity, pooled
—
Pages read by the fleet
these machines' own tally
—
Web pulled down
read, cleaned, filed
—
Shared card throughput
across — shared graphics cards
Built for the whole web
04 kinds of pageMost of the web is easy. The interesting part is the web that isn't — documents, app-built pages, and sites designed to keep automated tools out.
Ordinary web pages
Articles, docs, product pages, blogs — read, cleaned of clutter, and made searchable.
Documents
PDFs, Word files and slide decks are read page by page, not just linked to. Scanned pages are read too.
Modern web apps
Sites that assemble themselves in the browser are fully built first, so nothing is missed.
The hard cases
Pages that hide their content behind heavy scripting are handled as a last resort, rather than quietly dropped.
Real-time by default
fresh, not staleLegacy search
- Re-crawls on a schedule — days to weeks behind.
- Skips content that needs a real browser to render.
- Links to documents without reading them.
- Stores links, not understanding.
askFinz Web Index
- Indexed the moment a page is seen — real time.
- Reads modern and awkward pages properly.
- Reads documents all the way through.
- Stores full meaning, ready for AI to reason over.
Questions
04- What makes this different from a normal search index?
- Timing and depth. Pages become searchable the moment they're read rather than after a scheduled rebuild, and what's stored is the meaning of the page rather than a list of keywords — so an AI can reason over it instead of just matching words.
- Where does the computing power come from?
- From every surface askFinz runs on — the browser extension, the desktop app, devices running askFinz OS — plus a set of machines dedicated to the job, some lent by people with capacity to spare and some installed with partners.
- How fresh is it really?
- The live figures on this page are read from the running system every few seconds. Pages are indexed as they're read, and popular sources are revisited rather than being read once and forgotten.
- Can I get my own site in?
- Public pages are reached in the normal course of crawling. To be read directly and kept fresh with priority, become a verified domain partner.
Take part
Every reader makes it better.
Pick your way in.
Use askFinz and the extension reads alongside you. Own a site and you can decide how it's read, or feed it in directly as a partner. Have a computer with capacity to spare and it can take a share of the reading using only what it can spare — one line to start, one line to stop, nothing reformatted in between.