Use case · Finance and admin
Data out of documents, into your systems
A lot of office work is reading a document and typing parts of it somewhere else. We set up reading for the document types you handle most, with a person confirming whatever the model is not sure of.
Document queue
- Delivery note 5521, readScan
- Contract annex, two fields to confirmEmail
- Bank statement, March, readUpload
- Client intake form, readEmail
How you know you have this problem
People retype the same details from documents into two or three systems
Scanned papers pile up waiting to be entered
Finding one figure means opening PDFs one by one
Mistakes show up later, as wrong addresses, dates or amounts
What we build
How it works once it is running
- 01
Choose the documents
We start with one or two document types that take the most time and define exactly which fields you need from each.
- 02
Read with a model
Documents are read by an AI model that returns the fields in a fixed structure, with a note on anything unclear.
- 03
Review the uncertain ones
Clear documents pass through. Unclear ones wait in a short review queue where a person confirms or corrects the fields.
- 04
Send the data on
Confirmed data is written to your spreadsheet, CRM, ERP or archive, with a link back to the original document.
What you get
- A defined set of fields per document type
- A review queue for anything uncertain
- Data written to the systems you choose
- The original document stored and linked to the record
What we need from you
Sample documents of each type, including the awkward ones
The list of fields you use from them
Access to the systems where the data should go
Agreement on which documents may be processed by an external model
Questions about this use case
Can it read handwriting and poor scans?
Sometimes. Clean scans and digital PDFs read well; handwriting and photos taken at an angle are less reliable. We test on your worst examples first and route those to review if needed.
Where are our documents processed?
We agree this before building. Options range from a commercial model with a data processing agreement and no training on your data to a model hosted in the EU. Some documents may be excluded entirely.
What if we add a new document type later?
The setup is built to take new types: a new field definition, a few test documents and a destination. That is typical work for a care agreement.
More use cases for this team
Supplier invoices read, matched and ready to post
Incoming invoices are read, matched to purchase orders and sent for approval before they reach the books.
Open use case Business automationERP and business systemsOverdue invoices chased politely and on time
Reminders go out on a schedule from your open invoices, and the awkward cases go to a person.
Open use case Business automationAI integrationsThe monthly report, assembled for you
Figures from accounting, sales and operations pulled into one report on a schedule, with a written summary.
Open use caseShow us one process. We will show you what it could look like.
The analysis is free, takes about an hour of your time, and ends with a written summary you keep.
Prefer to read first? See examples