Operations · Document AI

Cutting manual invoice entry by 80% inside Inclined Labs' finance operations.

Project summary

Inclined Labs, a Swedish scale-up modernising for international expansion, was losing finance hours to manual invoice entry. An embedded engagement trained AI on the problem and wired it into the workflows, under governance the business could sign off.

The problem

Invoices arrived in every format a supplier could invent, and each one was keyed in by hand. Generic OCR failed on accuracy, and sending financial documents to external AI services failed on governance. The build had to be accurate, in-house and trusted by finance.

What we built

• Invoice recognition on Google Document AI, trained and tuned in house on the company's own document formats.
• Wired directly into the finance workflows, so extracted data lands where the team already works.
• Delivered under the company's governance model: sensitive financial data stays inside the business.

The result

Manual entry across finance operations is projected to drop by 80%. The model was trained on the company's own document formats rather than a generic OCR service, so accuracy holds across the invoices that actually arrive, and the extracted data lands directly where the finance team already works.

Related case studies