Case studySGCIE · REP · 2023–2024
ewen energyESCO · energy services

Two years of invoices into one REP,
without the retyping.

ewen energy used foraudits' automatic invoice extraction to compile the monthly consumption and costs of a two-year Execution and Progress Report under the SGCIE regime. What used to take weeks now takes days.

24
months of invoices structured
Weeks days
to compile the data
IE · CEE · IC
indicators fed straight from the invoices
+5,000
invoices extracted
+85,000
structured fields
94%
high-confidence fields
~30s
per invoice, with OCR
Zero
manual transcription
the context

A two-year REP starts with a stack of invoices.

ewen energy is a Portuguese ESCO that monitors facilities under the SGCIE regime. To submit an Execution and Progress Report, an industrial facility needs two full years of monthly consumption and costs, per energy source, taken from real invoices.

Done by hand, that means opening dozens of PDFs, copying values into a spreadsheet and reconciling everything. It's slow, error-prone and doesn't scale with the portfolio. What should be analysis becomes data entry.

SGCIEREP · bienniumindustrial facilityelectricity · gas · diesel
the result · monthly consumption rebuilt

24 months of electricity, extracted invoice by invoice.

20232024
14k10.5k7k3.5k0
J
A
J
O
J
A
J
O
Two-year total271 300 kWh
Equivalent312 tep
Total cost€ 41 320
our technology

OCR and AI trained to read energy invoices.

Every invoice runs through foraudits' extraction engine: OCR reads the document, the AI structures the fields and scores each one with a confidence level. The team only reviews what needs a human eye.

Any invoice, any supplier
Native or scanned PDFs, from dozens of suppliers, converted into a single data schema, with no per-supplier templates.
Confidence, field by field
Every extracted value comes with a confidence level. High-confidence fields are accepted automatically; the rest go to side-by-side review.
Fast, and at scale
Around 30 seconds per invoice, in batches of hundreds at a time. That's how more than 5,000 invoices became structured data.
how it ran

The same engine as always, on more than 5,000 invoices.

Invoices came in through the portal, OCR read each into structured data, and the ewen team reviewed side by side. Only low-confidence fields needed a human eye.

PDFs from dozens of suppliers, one data schema
Side-by-side review, original PDF always in view
Direct export into the REP calculation engine
Extraction confidence+85,000 fields
94%
High confidence, auto-accepted94%
Confirmed by the team5%
Corrected by hand1%
Each batch reviewed in under 2 hours
before and after

From weeks in a spreadsheet to days of analysis.

Before · email + Excel
Wk 1–2Chase and gather missing invoices by email
Wk 3Copy values PDF by PDF into the sheet
Wk 4Reconcile errors and redo the calculations
After · foraudits
Day 1Drop the invoices in the portal, auto-extraction
Day 1Side-by-side review of flagged fields
Day 2–3Indicators calculated and REP reviewed
Edgar Malato
ewen energy
“Extraction saved us the worst of the work: gathering and transcribing two years of invoices. We compiled the REP in days, with the team analysing instead of copying.”
Edgar Malato
CEO · ewen energy

Bring your next REP.

Book a demo and we'll run a batch of your energy invoices through collection, extraction and review.

Figures illustrative of a two-year SGCIE REP.