Pilardin now includes image processing and visual document understanding, so scanned pages, photos, and screenshots become usable data inside your workflows. Instead of retyping invoices or copying figures from photographed reports, teams can send the file to Pilardin and let it read the content directly.
The new capabilities cover two connected areas. Image processing handles practical preparation: rotating skewed scans, improving contrast on poor-quality photos, cropping borders, and splitting multi-page PDFs into individual pages. Visual document understanding then reads each page, recognizing printed and handwritten text, detecting tables and key-value pairs, and preserving layout details like columns and sections. The output is structured data that downstream steps can use, such as matching an invoice number to a purchase order or pulling line items into a spreadsheet.
In Pilardin, these steps work as part of your existing automations. A document arriving by email or uploaded to a shared folder can trigger a workflow that preprocesses the image, extracts the fields you need, and routes the result to the right person or system. If confidence is low on a particular field, human-in-the-loop review can flag it before anything is committed.
- Preprocessing for rotation, contrast, cropping, and page splitting
- Text extraction from printed and handwritten content
- Table and key-value detection with layout preserved
- Structured output ready for workflows and AI agents
This release is aimed at everyday document work: invoices, receipts, delivery notes, forms, and reports. It runs within your existing Pilardin environment, including on-premise deployments, so sensitive documents stay where your organization expects them. We will continue expanding language coverage and document types based on what teams use most.