🤖

Invoice OCR with artificial intelligence

JobIADoc — automatic digitisation connected to your ERP

Supplier invoices arrive in your mailbox, at the scanner or in your ERP’s capture screen, and JobIADoc reads them with Claude’s vision models: header, line items, amounts, VAT, withholdings and totals, with the lines adding up to the taxable base. It checks the supplier’s tax ID with the Spanish tax agency (AEAT), flags duplicates and leaves the document ready for the ERP to import, with no typing. 95-99% accuracy on any invoice format.

AI VisionMailboxScannerChecked tax IDMulti-companyREST API
Invoice received by email 09:01
AI reading the document... 09:01
Data extracted and tax ID checked 09:02
Ready for the ERP 09:02

Total time: 47 seconds

JobIADoc in action, with sample data

Real screenshots of the application

The invoice, read by the AI

JobIADoc OCR/AI screen: the invoice on the left and the extracted data on the right

Every document, with its status

List of captured documents with their status, amounts and totals

Detail with lines, shade and calibre

Invoice detail with issuer, recipient and lines with reference, shade and calibre

The history of every document

Document history: arrival, AI reading, tax ID check and saving

The mailbox, processed on its own

Mailbox emails with their invoices and the documents each one has produced

Home dashboard

Home dashboard with documents this month, issues and where documents come from

Capture from the ERP

Capture screen opened from the ERP with the invoice already read

Senders, companies and tax IDs

Settings for allowed senders, ERP companies and normalised tax IDs
👁️

AI Vision reading

Claude vision models that understand any invoice format: scans, phone photos and multi-page PDFs.

📄

Complete extraction

Header, lines with reference, shade, calibre and delivery note, VAT breakdown, withholdings and totals, with the lines adding up to the base.

🧾

Credit notes, currencies and special VAT

Credit notes and corrective invoices with the right sign, foreign-currency invoices, imports, reverse charge and intra-EU purchases.

📧

Mailbox

Checks the mailbox every few minutes: PDF, image and ZIP attachments, only from the senders you allow, and discards anything that is not an invoice.

🛡️

Checked tax ID, no duplicates

Checks the supplier’s tax ID with the AEAT (or VIES) and warns you if the invoice was already registered.

🔗

ERP integration

REST API and a capture screen that opens from the ERP. Native integration with ERP Jobers.

🖨️

Built-in scanner

Direct scanning from your multifunction printer via eSCL or WIA: scan and process in one step.

🕑

History of every document

What arrived, what the AI read, what was checked and who corrected what: full traceability for every invoice.

🏢

Multi-company

If you run several companies, each invoice is assigned automatically to the right ERP company from the customer’s tax ID.

Definition

What is AI OCR and why does it change invoice digitisation?

Traditional OCR (Optical Character Recognition) reads text from images using patterns and templates. To recognise an invoice, you have to configure where each field sits in each specific format. If the supplier changes its invoice, you have to configure it again.

AI Vision OCR needs no templates: it understands the document the way a person does. JobIADoc uses Claude’s vision models and a per-supplier learning layer that remembers where each piece of data is. It works with any invoice format, with low-quality scans, stamps or handwritten notes, and with long telephone, electricity, water or fuel-card invoices, keeping the lines that add up to the taxable base.

The result goes straight into ERP Jobers, ready to match against the purchase order and post to the accounts, with the supplier’s tax ID checked and no duplicates. No typing, no transcription errors.

Use cases

Documents JobIADoc processes automatically

📄
📄

Supplier invoices

Header, line items, amounts, VAT and withholdings. Credit notes, corrective invoices and foreign-currency invoices too.

📦
📦

Delivery notes and work reports

Delivery notes from suppliers and carriers and work reports, with the order of each line. Handwritten ones go for review.

📱
📱

Utility invoices

Telephone, electricity, water or fuel cards: the lines that add up to the base, without repeating the usage detail.

🛒
🛒

Customer orders

Orders that arrive by email, as a PDF or in the message text, with reference, model, colour and size on each line.

📧
📧

Email reception

A monitored mailbox: invoices that reach your inbox, even inside a ZIP, are processed automatically.

🖨️
🖨️

Direct scanning

eSCL and WIA support: scan and process in one step from the multifunction printer.

ROI

How much time you really save

95-99%

accuracy extracting key fields (amounts, VAT, totals)

4-8h

saved every week by the admin team (200-500 invoices/month)

2-4 months

typical ROI from go-live

Frequently asked questions

What people ask us most about JobIADoc

How accurate is AI OCR compared with traditional OCR?

JobIADoc uses Claude’s vision models instead of traditional pattern-based OCR. That means 95-99% accuracy on key data (amounts, VAT, totals) on any invoice format, even with poor scan quality, stamps, handwritten notes or unusual layouts. Traditional OCR depends on templates and fails when the supplier changes the design.

How does it learn to read each supplier’s invoices?

When a field is corrected or marked on an invoice, the system saves that supplier’s mapping: where each field is and what it looks like. The next invoices from the same supplier are read following that pattern. There are no templates to set up by hand.

Does it integrate with any ERP or only with ERP Jobers?

With ERP Jobers the integration is native: the extracted data reaches the purchasing module ready to match against the order and post to the accounts. For other ERPs there is a REST API with a key per company and a capture screen that the ERP opens with its own reference. Ask us about your specific case.

What happens with invoices that cannot be read properly?

The document is saved anyway, marked as an issue with the reason in plain sight: tax ID not found at the AEAT, duplicate invoice, lines that do not add up or a handwritten delivery note. Your team only reviews those documents instead of typing 100% of them, and a corrected tax ID is remembered for future invoices.

How much time do you really save?

In companies that were processing 200-500 invoices a month by hand, the saving measured after 3 months of use has been 4-8 hours a week for the admin team. In companies with larger volumes (1,000+ invoices a month) the saving is proportionally greater. Typical ROI: 2-4 months.

Can it be installed on our own server?

Yes. JobIADoc runs in the Jobers cloud or installed on your company’s server, with its own database. The mailbox, the scanner and the ERP integration work the same way in both cases.

Is it secure? What happens to my invoice data?

Invoices are processed on our own infrastructure and are not kept by third-party providers. The AI models receive only the content needed for the extraction and are not used for training. GDPR compliance, hosting in Spain and encryption in transit and at rest.

How much time does your team spend typing invoices?

A personalised demo with real invoices from your suppliers. No obligation.

✦
¡Hola! Soy tu Comercial Mágico
Tu asistente IA de Jobers