The Client: Spendesk

Spendesk is the all‑in‑one spend management platform that helps businesses save time and money by unifying company spending.

By combining everyday technologies, built‑in automation, procurement tools, and an intuitive approval process, Spendesk empowers employees to spend efficiently while offering finance teams full visibility into expenditures.

With the launch of its Procure‑to‑Pay solution, Spendesk became the first European platform to fully integrate procurement and spend management for companies of up to 1,000 employees.

The Challenge

Before partnering with Mindee, Spendesk had tested other OCR solutions but encountered limitations in key areas:

• Speed & Scale: Spendesk needed a fast, reliable way to extract data from thousands of receipts and invoices each day.
• Accuracy: Errors in OCR output led to manual fixes and reconciliation issues downstream.
• Document Versatility: Handling both receipts and invoices within the same workflow required a flexible, high-performing solution.

The Solution: Mindee OCR Integration

Spendesk developed a lightweight OCR microservice written in Python that makes HTTP requests to Mindee’s OCR API.

This service is called by Spendesk’s backend, built in TypeScript (Node.js). When a document is uploaded via the Spendesk platform, the backend triggers the OCR microservice, which then forwards the document to Mindee for data extraction.

Thanks to Mindee’s clear documentation and responsive support, the integration was completed within a few weeks and now handles more than 13,000 OCR requests per day.

Key Features Leveraged

• Invoice OCR: Extracts essential fields like invoice number, due date, amounts, VAT rates, and currency.
• Financial Document OCR: Automatically detects and processes both receipts and invoices in a single endpoint—reducing complexity in Spendesk’s workflows.
• High Throughput: Capable of consistently processing 8,000 financial documents and 5,000 invoices daily.
• Flexible Integration: Mindee’s API was integrated via a custom microservice, enabling fast adaptation to Spendesk’s existing stack.

Integration and Impact

Mindee’s OCR solution had a measurable impact on Spendesk’s product and internal workflows:

• Operational Scale: Over 13,000 daily document extractions without performance bottlenecks.
• Speed of Extraction: Mindee’s fast response times were a standout feature, supporting real-time processing needs.
• Workflow Automation: By pre-filling document fields, Spendesk significantly reduced the need for manual data entry—supporting smoother approval and reconciliation flows.
• Full Platform Benefit: 100% of Spendesk users benefit from Mindee’s OCR capabilities, making it foundational to several features in the platform.

Partnership Approach

The collaboration between Spendesk and Mindee goes beyond technology.

Mindee’s team has remained attentive and responsive, partnering closely with Spendesk to improve core extraction fields such as:

• Amount and currency (particularly for Irish invoices)
• VAT rate, amount, and base
• Document number and due date

Looking Ahead

Spendesk continues to explore new ways to enhance its document processing capabilities with Mindee. Future areas of focus include:

• Field Accuracy Improvements: Fine-tuning precision for regional invoice formats.
• Unified API Logic: Consolidating receipt and invoice extraction endpoints for simplified integration.
• New Use Cases: Investigating additional document types and automation opportunities.