Identify Entidade, Referência & Amount from a Multibanco Payment Slip
Manually copying Entidade, Referência, and amount from a fading thermal slip into your accounting software takes 2 minutes per receipt — this does it in 5 seconds from a phone photo.
5-10s per slip · Handles faded thermal paper
Three Numbers That Matter on a Multibanco Slip
A Multibanco slip carries only three data fields — no merchant name, no item list. With Custom Column Extraction, you type the field names and the AI finds each value by understanding what it means, not where it sits.
A 5-digit code identifying the receiving institution — EDP, AT (tax authority), or a municipality. It is not the merchant name but a SIBS registry number.
A 9-digit payment reference with a check-digit that ties the payment back to a specific invoice — the most critical field for reconciliation.
The payment value in EUR. Portuguese formatting uses a comma as decimal separator (€25,50). The AI normalises it to standard decimal format in the output.
These three column names are all you need. The AI finds them on any Multibanco slip — faded, curled, or fresh.
This Isn't a Screenshot — It's Fading Thermal Paper, and That Changes Everything
Every other payment node in this series covers clean digital screenshots. Multibanco is the exception: a physical thermal slip photographed with a phone camera. Traditional OCR was built for flatbed scans of white paper — it breaks on the exact conditions a real Multibanco slip brings.
Why Multibanco Slips Break Traditional Tools
Thermal paper fades — and keeps fading. Multibanco slips are printed on heat-sensitive paper that degrades from the moment it leaves the ATM. A slip that was crisp at the till can be unreadable three weeks later in a wallet. Template OCR expects clean, stable characters — fading breaks character recognition at the pixel level.
Phone photos add glare, shadow, and angle distortion. A photographed slip is not a clean scan. The camera angle skews the text plane. Shadows from overhead lighting darken the image. Glare from the thermal paper surface washes out parts of the print. Traditional OCR reads individual characters — it has no concept of "that dark patch is a shadow, not part of the number."
No merchant name — just a code that template tools don't understand. A Multibanco slip does not say "EDP" or "Finanças." It says "Entidade: 10083." Template-based receipt scanners look for vendor names and line items — neither exists here. They either fail silently or output garbage.
How Semantic Extraction Reads Slips by Meaning
The AI reads the whole scene — not isolated characters. A visual language model processes the photograph as a complete image: paper texture, crease running through a character, shadow across the print. It reconstructs the text from visual context rather than from individual pixel shapes. This is why it handles faded thermal print that breaks traditional OCR.
You name the fields — AI separates Entidade from Referência by context. Type "Entidade (Entity)", "Referência (Reference)", "Montante (Amount)". The AI understands that a 5-digit number next to "Entidade:" is the institution code, and the 9-digit number further down is the payment reference. It doesn't need to know where each field sits — it reads the labels.
Portuguese number formatting is normalised automatically. A comma-formatted "25,50" on the slip becomes "25.50" in the output spreadsheet. The AI recognises the Portuguese decimal convention during extraction — you don't have to manually convert currencies or reformat amounts.
From a Phone Photo of a Thermal Slip to One Spreadsheet Row
Photograph the Slip
You have a stack of Multibanco payment slips — some from the ATM, some printed at a store POS. They are on thermal paper, some weeks old, the edges slightly curled. Take a photo of each one with your phone. Lay the slip flat on a table, shoot from directly above in even light. JPG or PNG, no pre-processing needed.
Name the Three Columns
Type Entidade (Entity), Referência (Reference), Montante (Amount). Three column names, one set, works for every slip in the batch. The AI reads each photo individually — it understands that the 5-digit code near "Entidade:" is one field and the 9-digit number near "Ref." is another, regardless of which ATM printed the slip or how faded the print is.
Download Your Reconciliation Table
Processing takes 5-10 seconds per slip. The output is one clean XLSX or CSV file: one row per slip, three columns exactly as you named them. A stack of 20 end-of-month slips turns into a 20-row reconciliation table ready for Primavera, PHC, Sage Portugal, or your own accounts spreadsheet — 18x faster than copying each number by hand.
What a Photo of a Multibanco Slip Can and Can't Do
Understanding the tool's boundaries helps you get the best results from photographed thermal slips.
When It Works Best
Slips photographed flat from above in even light. The AI reads the visual scene, not individual character shapes — moderate fading and curling are handled well.
Standard SIBS-format slips from any Portuguese bank or ATM. The same three column names work across the entire Multibanco network.
Batch processing for month-end reconciliation. Upload a stack of slips together and get one merged spreadsheet, no per-slip handling needed.
When to Be Cautious
Completely faded thermal paper with no visible print. If the thermal coating has fully degraded and no text is visible to the eye, no extraction method can recover data that is no longer physically on the paper.
Deep creases or tears through numeric digits. A sharp fold through an "8" or a "3" can break character continuity. Moderate creasing is handled well, but slips with physical damage through the Entidade or Referência numbers should be spot-checked.
Frequently Asked Questions
Can I extract data from a faded Multibanco slip that's been in my wallet for weeks?
If the characters are still faintly visible to your eye when you look at the phone screen, a visual AI model can usually read them — it detects text at the pixel level even when contrast is low. If the thermal paper has turned completely dark or the print has vanished entirely, no extraction method can recover data that is no longer physically present on the paper. The earlier you photograph the slip after receiving it, the better the results.
The slip shows "Entidade: 12345" — can AI tell me which company I paid?
The AI extracts what is actually printed on the slip. Standard Multibanco payment confirmations do not display the merchant name — only the 5-digit Entidade code. The AI will extract that code accurately, but it cannot invent a merchant name that was never printed. To identify the institution, you can cross-reference the Entidade code against the SIBS entity registry or your own accounting records. Portuguese accounting software like Primavera BSS, PHC Software, Sage Portugal, Moloni, and InvoiceXpress can match the Entidade + Referência pair to the correct invoice automatically during reconciliation.
Can I process 20 Multibanco slip photos at once into one spreadsheet?
Yes. The tool is batch-first by design. Upload multiple photos of Multibanco payment slips simultaneously, type your three column names once (Entidade, Referência, Montante), and run extraction. The output is a single merged XLSX or CSV with one row per slip — a 20-slip stack turns into 20 rows ready for reconciliation. This is by far the most common use case for Portuguese businesses doing month-end accounts receivable matching.
The slip uses Portuguese decimal formatting like "25,50" — will the output use normal decimals?
Yes. The AI's post-processing recognises Portuguese decimal conventions and normalises them during extraction. A comma-formatted amount like "25,50" on the slip becomes "25.50" in the output spreadsheet, matching the standard spreadsheet number format. The original formatting is visible in the extracted raw data, but the normalised value goes into your table — no manual reformatting needed.
Can I use a screenshot from online banking instead of a photo of the paper slip?
Yes — the visual AI processes any visual input the same way. Whether you photograph the paper slip or take a screenshot of the Multibanco confirmation page from your bank's homebanking portal, the same three-column extraction works. The digital screenshot produces cleaner results since there is no thermal fading or camera distortion to contend with. The paper slip scenario is the harder case, but both are supported.
Deep dive: How to extract the Entidade, Referência, and amount from faded Multibanco thermal paper slips
See all supported payment apps and how they compare: Payment Screenshots Hub