Leverage with AIDocuments and books
Invoices and receipts copied into a spreadsheet one at a time
A PDF or a photo becomes a row: supplier, date, total, line items. The unclear ones stay with a person.
Who this is for Businesses that take dozens of supplier invoices and receipts a month, where the books start with retyping.
The problem
Invoices retyped by hand
The document arrives by email, through a messaging app, as a photo on somebody's phone, or on paper. Someone opens it, finds the supplier, the date and the total, retypes them into a spreadsheet or the accounting software and files the original somewhere it will hopefully be found. None of it is hard, there is simply a lot of it, and it piles up until the end of the month when there is least time.
You will know it by
- The end of the month is an evening spent with a pile of PDFs.
- The same total gets entered twice: once in a spreadsheet, once in the books.
- Receipts sit in a phone's photo roll until somebody deals with them.
- The bookkeeper asks for a document nobody can find.
Possible solutions
Not every road leads to me
Your accounting software's own scanner
Most modern accounting packages read documents already.
Where it stands The first thing worth checking, because you already pay for it. Sometimes the right answer is to switch on what you have.
A bookkeeper who takes the filing too
You hand over the pile and pay for the sorting.
Where it stands The calmest route when the price suits. Your numbers still reach you only after somebody has keyed them in.
Reading them into rows
The document becomes a row with supplier, date, total and line items, with a person checking the unclear ones.
Where it stands This is the one I build. Worth it when the volume is real, the documents arrive from several places, and the built-in scanner did not cope.
What I set up
One place documents land: a dedicated email address, a folder in the cloud, or a photo taken on a phone. From each one the supplier, date, total and line items are read out and land in your spreadsheet or your accounting software. The original is renamed and filed where your bookkeeper will also find it. Anything the reading was not sure about goes into a separate list for a person to confirm.
How it goes in
- We pick the one place every document goes from now on.
- We check first whether your accounting software already does this.
- The reading is tested on your own documents from last month.
- We agree who checks the unclear ones, and how often.
The limit Unclear, handwritten or badly photographed documents keep a step where a person checks them. Whether the books are right stays your bookkeeper's responsibility, not the scanner's.
The work It depends on how many shapes arrive. One steady set of suppliers goes quickly; twenty different layouts takes longer.
The prices on this site are final.
What it brings
A month end without the evening pile
Documents are dealt with as they arrive rather than when it is already late.
Numbers you can look at now
Spend by supplier or category is visible without waiting for the month to close.
The original is findable
The file is named so that both you and your bookkeeper can find it.
How you check it yourself
Note how long last month's documents took to process and how many had to be hunted for twice. Repeat after two months with the same two numbers.
Is this your job?
Tell me how this runs at your place now. On a 30 minute call I will say whether it is worth automating, where I would start and what it needs.
Book a 30 minute callThe call is free, with no obligation and nothing to sign.
Not sure which job is first for you? The free assessment says where the hours go. Start the assessment
Common questions
Does this replace the bookkeeper?
No. It removes the retyping, not the accounting. Your bookkeeper still answers for the entries being right, they simply start from data that is already in.
Do our invoices leave the company?
It depends on the design, and that is decided before work starts. Some of this runs entirely inside your own environment. Where an external model is used, it is named in the contract as a sub-processor.