Stuck, puzzled,
or found a bug?

No. Reading, browsing, searching and previewing the proposed tidy-up all happen without touching a single file on disk. Files move exactly once — when you commit a layout you have looked at — and the commit sheet states exactly what will happen before it does. Every commit is reversible.

Only what you tell it to, and only into your Recycle Bin.

Two things are set to be cleared out when you reorganise in place: extra copies of files you already have elsewhere, and data that programs made for themselves. The commit sheet shows both counts, lets you keep either, and asks about the deletion on its own before anything happens. What goes, goes to your Recycle Bin — or your cloud account’s trash if the files live there — so you restore any of it exactly the way you restore anything else, and the record Ontiqo writes lists every file by name.

Choose to build the organised copy somewhere else instead and nothing is deleted at all: those files are simply not carried across, and everything you started with stays where it is.

Yes. All processing happens on your own computer and none of it needs a connection. The internet is involved only when you connect a cloud source, switch on an optional lookup, or activate a work licence.

The full text of documents and spreadsheets. Scanned pages, read as if they were typed. Photos, looked at and described — including any words visible in them. Speech in audio and video, transcribed into searchable text. Archives, discs and programs, identified for what they are.

Some files cannot be read — password-protected, damaged, or in formats nothing opens. Those are named, counted and shown on their own shelf, never silently dropped.

It depends on the size of the collection and the machine. A few thousand files is hours of background reading; tens of thousands is days. Ontiqo shows its pace and an estimate from its actual recent speed — and you do not wait for the end, because files become searchable as they are read.

One button, top right. It also pauses itself out of courtesy — when you are presenting, in a full-screen app, or on battery — and catches up when the machine is idle.

Search tools — including Windows' own, which on recent machines can find a photo from a description of it — answer "where is that one file?". They are good at it, and they leave your collection exactly as it was. Ontiqo answers a different question: what is in here, and how should it be arranged?

Sorters and renamers tidy a folder into categories they arrived with, or into a scheme you describe to them first. Ontiqo has no preset categories and does not ask you to design a filing system: it reads the collection and derives the structure from what is actually in it, which is why a photo archive and a paperwork drawer come out shaped completely differently.

Three other practical differences: Ontiqo reads what was said in recordings and what is in photographs, not just documents; it works on whole collections of tens of thousands of files rather than in batches; and it proposes the entire arrangement for you to approve before anything moves, with every commit reversible. Searching your files afterwards comes free, because everything has already been read.

Yes — this is a design rule, not a happy accident. Ontiqo derives structure from whatever the collection actually is: a photo archive gets groupings a photo archive supports, a music library gets musical ones. There are no preset categories to mismatch your files.

It is best in English today. Documents and recordings in other languages are handled with mixed results — that is a limitation, not a footnote.

A single organising engine, built around models we have trained and tuned for the job, running entirely on your own computer. At its centre is a language model of roughly eight billion parameters that we have fine-tuned for one task: reading a file in full and producing a structured record of it — what it is, what it is about, and the facts a filing decision turns on — in a fixed shape, so nothing is left to free-form guessing. Around it, the same kind of record is produced for files that are not text: a vision model describes photos and video frames, speech recognition transcribes recordings, OCR reads scanned pages and the words visible in images, and archives, discs and programs are identified from their own structure. Different readers, one pipeline, one record per file — with a representation of each file's meaning kept alongside, so files can be compared by what they are about as well as by their facts.

It derives them from the collection rather than starting from a template. Once enough has been read, the records are analysed together: a tuned model proposes candidate ways of grouping — the recurring kinds, subjects and groupings the collection actually contains — and then a statistical stage judges every candidate against the collection itself, keeping only the groupings that are genuinely shared across many files, that add something the others do not, and that hold up when the analysis is repeated. What survives is the collection's own schema: a small set of dimensions with a consistent vocabulary and a natural order from broad to specific. Because it starts from the collection rather than a template, a photo archive and a paperwork drawer come out organised completely differently.

From the evidence the file itself carries — its record, its meaning, its metadata — weighed by how trustworthy each source is, so what a file says about itself always outranks the folder it happened to be sitting in. Files that already belong together as a set are recognised as a set and kept together. Duplicates are matched by their contents, not their names. And anything the evidence cannot settle is shelved and counted rather than forced into a folder to make the tree look finished.

Because it reads. A model reading a document in full costs seconds of compute per file, transcribing an hour of audio costs minutes, and a large collection multiplies that by tens of thousands. Ontiqo spreads the work across your processor and, when you have one, your graphics card — giving each kind of work a share proportional to how much of it is left, so nothing sits idle while something else runs for hours — keeps its models loaded rather than reloading them, and holds the whole thing at low priority so the machine stays yours. What it never does is trade the reading down for speed: a smaller collection just finishes sooner.

They are named on the About page inside the program, with their licences, and they change as the program improves — so we describe them here by role rather than by name. Every one of them runs on your machine and ships inside the download; none is a cloud service.

No. Reading, understanding and organising all happen locally, and your file content is never sent to us — that is written into the licence, not just this page. The things that can go online are named and narrow: the optional lookups you switch on (each states what it sends, beside its switch), sign-in to a cloud source you connect, and licence activation on the work edition. None of them carry your files.

No. Searches and questions are answered on your machine, from your index. There is no account, and nothing reports them to us — we could not look them up if you asked us to.

Beside the program, on your machine. It holds what Ontiqo read and worked out — extracted text, descriptions, groupings — so it can answer without re-reading everything. Your original files are never copied into it, and removing Ontiqo removes the index with it.

Connecting a source signs you in with that provider, on your own account, so Ontiqo can fetch your files to your machine and read them there. That conversation is between your machine and your provider — we are not in it.

Yes — genuinely. As many of your own computers as you like, no time limit, nothing switched off, and no locked "pro" version. If you are using Ontiqo for your own things, you do not need a licence and should not buy one.

It doesn't check, and it doesn't try to. The free program and the licensed program are the same program — no nag screen, no watermark, no held-back feature, and nothing that reports on you. A work licence is an honest arrangement: if Ontiqo is doing a job for a business or organisation, the organisation pays for the right to use it. That is the whole mechanism.

When Ontiqo is used for a business, an employer, a not-for-profit, an educational institution, or a government or public body. Every computer it runs on for work has to be covered. The software is exactly the same either way — what you are buying is the right to use it at work. Details and prices are here.

A single Work Edition covers two — a desktop and a laptop, say. The 5 Pack is five licences covering up to ten computers in total, and the Organisation licence covers unlimited computers within one organisation. If a machine is permanently retired, its activation can move to a replacement.

Refunds are 30 days, no questions asked — for any reason, including that you bought a licence you did not need. Open a ticket above, or email from the address you bought with, and it will be refunded in full. After 30 days, contact us anyway; where something does not work as described we will still put it right. Checkout, receipts and payment will be handled by a merchant of record, so your receipt will come from them rather than from us. They will be named here before any licence goes on sale.

Only from ontiqo.com, or the release page this site links to. Ontiqo is distributed solely by us; nobody else is authorised to host, mirror or bundle it, and the licence agreement forbids redistributing it. A copy from anywhere else is unauthorised — it may have been altered, and it is not covered by our support. The download page shows the size and checksum of the genuine file so you can check what you got. If you see Ontiqo offered elsewhere, please tell us.

Nothing of Ontiqo — no service, no system components, no leftovers. The one thing that survives is yours, not ours: the plain-language undo record each commit writes into your own Documents folder, kept there precisely so it outlives the program.

Yes. Here is the shape of it. The undo record lives in your own Documents folder, so uninstalling never deletes it, and beside it sits a double-click undo that needs nothing installed: for files on your machine, run it and every file goes back where it was, with a log of what it did. For files in a cloud account, undoing needs your sign-in, which a script cannot carry — download Ontiqo again (it is free), point it at the same record, and it does the rest. The record is also plain text a person can read: it lists every file's old and new place.

And if you committed into a new folder, your originals were never modified in the first place — the new tree is built as copies.

Every step is journalled before it happens, so an interrupted commit either resumes cleanly or rolls back cleanly — and the rollback can be verified against the record, not taken on faith.