The AI indexing
Scanning turns paper into images. Indexing is what makes those images findable, and it is normally the most expensive line on a scanning invoice because a person sits and types it. We do it with software.
The difference, concretely
What you usually get back
One file per sheet of paper, named by a counter. Searchable only if you already know which file to open.
Scan0001.pdfScan0002.pdfScan0003.pdfScan0004.pdfWhat we send back
One file per real document, named from what is inside it, with metadata attached so you can filter and sort.
2019-03-14 Invoice 10482 - Coates Hire.pdf2019-03-18 Work Order 59697 - 14 Marine Pde.pdf2019-04-02 Contract - Brady Constructions.pdfHow it works
It finds where documents begin and end
A box is not a stack of unrelated pages, it is a stack of documents that happen to be several pages long. The AI reads the pages in order and decides where one document stops and the next starts, so a six page contract comes back as one file rather than six.
It works out what each document is
Invoice, work order, contract, certificate, statement, correspondence. Once it knows the type, it knows which fields are worth pulling out, and those become the file name and the metadata.
It captures the fields that matter to you
Up to five per document, included in the price. Which five is your decision, and it is usually obvious once we see a box:
- Date on the document, not the date it was scanned
- Who it is from or about
- Reference numbers: invoice, work order, matter, patient, job
- Property, site or address
- Amount, where there is one
It admits when it cannot read something
This is the part we care most about. A faded fax or a page of handwriting will sometimes defeat it. When that happens the document is still filed and still full-text searchable, but it is flagged for a person to check rather than given a confident, wrong answer. A wrong label is worse than no label, because you stop trusting all of them.
If the AI is unavailable, a run stops before filing anything rather than processing a batch badly. Your boxes stay where they are and the run picks up again afterwards.
Where it ends up
Indexing is only worth paying for if the metadata survives the trip. A PDF with a good file name in a folder on a drive is still a folder on a drive.
SharePoint
Our usual recommendation, and what we know best. Each document becomes a list item with real metadata columns, so your staff filter by date range, by client, by reference number, in the tool they already have open. No new software, no new login, and it is searchable from the Microsoft 365 search bar.
Your own system
If you run a CRM, a practice management system or a job management tool with an API, we can push documents straight into it against the right record. We are a software development company, so this is ordinary work for us rather than a special project. It is quoted separately because it depends on the system.
Somewhere simpler
A cloud drive, a network share or an encrypted USB, with the metadata as a spreadsheet alongside. Less useful, but it is your call and sometimes it is genuinely all that is needed.
Why we can include this
Per-field indexing is normally billed at 50c to $2 a document because somebody is reading each one and typing the answer into a form. On a box of two hundred documents that is real money and real time, which is why most jobs skip it and end up as Scan0001.
Our own developers, here on the Gold Coast, built this pipeline for a client who needed their work orders read, named and filed into SharePoint automatically. It already runs in production. Pointing it at someone else's archive costs us a fraction of what a person doing the same job would, so we include it rather than charging for it, and compete on what you get back instead of on cents per page.
Try it on one box
Send us a box, or let us collect one. We will scan and index it free, and you will see exactly what the file names and metadata look like on your own documents before deciding anything.
