Start a project
Selected work
Case study 02 / AI · Web Application

Invoicely.

From invoice documents
to usable information.

AI-assisted extraction brings invoice information into a structured workflow, with context for review and a place to manage the result.

  • Next.js
  • Supabase
  • AI-assisted extraction
The transformationConceptual workflow · not a product screenshot
  1. 01DocumentThe original invoice
  2. 02AI extractionRelevant information identified
  3. 03Structured dataIndividual invoice fields
  4. 04Invoice workflowReview and manage invoices
Document → Structured result

The information is there.
Give it structure.

Invoices contain useful business information, but arrive as documents. Reading each one and entering its contents manually separates the document from the data a workflow needs.

Invoicely processes an invoice and extracts relevant information into individual fields, making the result available within the application.

Extracted invoice fieldsInformation model

Invoice identity

  • Invoice number
  • Invoice date

Parties

  • Supplier
  • Customer

Amounts

  • VAT
  • Subtotal
  • Total
Invoicely invoice extractor showing the uploaded invoice and extracted supplier, customer, VAT and total fields.
Invoice detail / extracted information

View the full screenshot. Choose Actual size to inspect the details.

01The individual invoice view: the place for structured information from a real extraction.
Extraction → Review

Useful output.
Visible context.

AI output should not be presented as unquestionable. Confidence, warnings and raw text provide context around the extracted fields.

Invoice 1047 detail with its confidence indicator, structured invoice fields and expanded raw extracted text.
Confidence / warnings / raw text

View the full screenshot. Choose Actual size to inspect the details.

02A real capture of the review context accompanying an invoice extraction.
01Confidence
A confidence field adds context to the extraction. It is not a guarantee that every value is correct.
02Warnings
Warnings accompany the extracted information, helping keep potential issues visible during review.
03Raw text
The extracted raw text remains part of the result, alongside the structured invoice fields.
Structured result → Application

More than
an extraction.

The product includes a landing page and login, with authentication and Google OAuth implemented. Inside the application, a dashboard, invoices, invoice history, individual invoice pages and a billing area give the workflow a home.

Invoicely dashboard showing invoice usage, document tools and recent invoices.
Dashboard

View the full screenshot. Choose Actual size to inspect the details.

03The authenticated application dashboard.
Invoicely invoice history showing invoice summaries and a table of previously extracted invoices.
Invoice list / history

View the full screenshot. Choose Actual size to inspect the details.

04Invoices and their history within the application.
Implemented application areas
  • Landing page
  • Login
  • Dashboard
  • Invoices
  • Invoice history
  • Invoice detail
  • Billing
Product foundation

A defined starting point.
A connected stack.

Implemented Free plan
10invoices
per month

The Free plan has an implemented monthly limit of 10 invoices.

Web application
Next.js App Router
Authentication
Supabase authentication Google OAuth
Invoice data
PostgreSQL Supabase database
Document processing
AI-assisted invoice extraction
Start a conversation

A document workflow
worth improving?

Start a project Back to selected work