How to extract invoice data from PDF to Excel automatically (no code)
Every month the same job: invoices arrive as PDFs — some typed, some scanned, a few photographed on a phone — and someone opens each one and copies the supplier, the number, the date and the total into a spreadsheet. Twenty invoices is an hour. Two hundred is a day, and by invoice one-fifty the mistakes start.
This is how to make that job run by itself, without writing code — and, if you want, without sending your invoices to anyone else's servers. It takes about fifteen minutes to set up the first time.
What "automatically" has to mean for invoices
Any tool can pull text out of a PDF. The hard part is the three things that go wrong in real accounting work:
- Scans and photos. Half of what arrives is a picture, not text. The tool has to read the picture — not fail quietly and leave an empty row.
- Numbers that don't add up. Line items that don't sum to the total, a total ten times higher than that supplier usually charges. A tool that copies these into Excel faithfully has automated the mistake.
- Who checks? A row nobody looked at is a row nobody trusts. There has to be a place where the doubtful ones wait for a person.
The setup below does all three. It uses AIK Platform, which runs on your own PC or server; the same steps apply to any tool that offers them.
Step 1 — say what you want out of each invoice
Open Document Types and pick the built-in Invoice type, or press + New type to make your own. A type is just a list of fields: vendor, invoice number, date, total, VAT, line items. Each field has a kind (text, number, date, list) and, if you want, a rule: required, a minimum or maximum, a pattern for the invoice number.
That is the whole "programming". You are describing the columns of the spreadsheet you already keep.
Step 2 — say where the result goes
On the same page, choose what happens with each record: append a row to an Excel file (say Invoices.xlsx), save it in the platform's own records, e-mail you a summary, or all three. If you already have accounting software with an API or a database, the same record can be posted there instead — but for most people Excel is the point, so start there.
Tick require approval if you want nothing written until a person has looked. For the first month this is the right setting.
Step 3 — point it at the invoices and press Run
Put the PDFs in a folder — a folder on the machine, a shared drive, or the platform's own upload page — and press Run. If invoices arrive by e-mail, the Mail agent can pull the attachments into that folder first; if they arrive on paper, scan them into it.
Now each file is read by an AI model. A typed PDF is read as text. A scan or a phone photo is shown to the model as an image, so a crooked photo of a receipt comes out with the same fields as a clean PDF. You choose the model, and this is a real choice: with a cloud model (OpenAI, Anthropic, Google) each invoice is sent to that provider to be read — fast and accurate, but your documents do go to them, under their terms. With a free local model through Ollama, running on your own machine, no invoice ever leaves your network — slower on an ordinary PC, and the answer to the GDPR question before it is asked. Many firms use a local model for invoices and a cloud one for everything else.
Step 4 — look at what it flagged, not at everything
This is where the time saving actually comes from. Every record gets checked before it is written:
- Required fields missing — the record is listed as failed, with the reason.
- The maths — line items are summed and compared with the total (allowing for VAT on top of the items). A clear mismatch marks the record pending.
- A supplier you have never paid — flagged, because a first invoice from a new name is exactly what an invoice-fraud attempt looks like.
- An amount far from that supplier's usual — flagged the same way.
- The VAT itself — for Bulgarian invoices, that VAT is 20 %, 9 % or 0 % of the net and that net + VAT is the total; a wrong rate or a wrong sum is flagged with the numbers in the reason.
- The supplier's company number — the ЕИК/Булстат (or the BG VAT number) is verified with its checksum, so a misread digit is caught before anyone pays.
- A PDF's digital signature, if it has one — checked, and reported.
So instead of reading two hundred invoices you read the six that were flagged, press Approve on the ones that are fine, and the Excel file fills itself. Nothing doubtful is written without a person having seen it.
Step 5 — make it run on its own
Set a schedule on the type — every night, every hour, or the moment a new file appears — and the job is gone from your day. Each run leaves a record of what it did, and the Analytics page counts the hours it saved, so you can tell your boss a number rather than a feeling.
What this costs
The setup is fifteen minutes once. Per invoice, a cloud model charges a fraction of a cent; a local model charges nothing. On a modest office PC a scanned page takes a few seconds to read.
Try it
AIK Platform has a free tier — start here, or read how the document pipeline works. For an accounting firm that wants it installed on its own server, with the first document types set up, get in touch.
More from AIK
- A self-hosted alternative to cloud automation tools for AI workflows
- How to run AI agents on your own server with Ollama and AIK
- EU AI Act Article 50: how to mark AI-generated documents and calls
- How to build a phone answering AI assistant
- Invoice fraud: how a swapped IBAN gets caught before payment
- What is a "process audit" for automation, and what does it tell you
- How much does an AI agent cost to run per month
- RPA vs AI agents: which processes belong to which
- How to give Claude access to your own automation platform (MCP)
- Как да автоматизирам въвеждането на фактури в счетоводството
- AI асистент, който вдига телефона на български — как работи
- Как да си направя чатбот за сайта без програмиране
- Автоматични отговори на имейли с изкуствен интелект — безопасно ли е
- Какво е on-premise AI и защо счетоводна кантора би го предпочела
- Регламент за изкуствения интелект (AI Act) — какво трябва да пише на документите
- Как да проверя дали PDF фактура е променяна
- Алтернатива на UiPath за малки процеси
- AI за счетоводители: какво поема изкуственият интелект и какво остава за вас
- Автоматизация на счетоводството: откъде да започне малката фирма
- Изкуствен интелект за малък бизнес: пет задачи, които да дадете на AI още този месец
- Автоматизация на офис процеси: как да намерите кое си струва
- AI accounts payable automation: from inbox to approved invoice
- GDPR compliant AI automation: five questions to ask before an AI touches personal data
- AI agents for small business: what they really do, and what to give them first
- Ready-made AI agents: start with one sentence, not a blank canvas
- Готови AI агенти: започнете с едно изречение, не с празен лист
- An AI operating system for business: one place where AI work is asked for, done and checked
- AI операционна система за бизнеса: едно място, където AI работата се поръчва, върши и проверява
- Model-agnostic AI agents: OpenAI, Claude, Gemini and a local model in one platform
- AI с OpenAI, Claude, Gemini и локален модел — в една платформа, без абонамент на заявка
- Custom tools for AI agents: when the ready-made ones are not enough
- AI агент със собствени инструменти: когато готовите не стигат
- Secure AI agents: sandbox, permissions and a human in the loop
- Сигурни AI агенти: изолация, права и човек, който одобрява
- AI agent integrations: connect the tools you already use — and add your own
- Интеграции на AI агенти: свържете инструментите, които вече ползвате — и добавете свои
- Intelligent document processing without an enterprise project
- Интелигентна обработка на документи с AI — без голям проект
- Одит на процеси за автоматизация: кои процеси да автоматизирате първо
- A WhatsApp chatbot for business that also answers the phone and your website
- Как работи обработката на фактури в AIK — стъпките, проверките и предпазителите
- Как AI асистент записва часове във Viber — и кога предава на човек