Skip to content

AIIDA Release notes

2021-09-15

  • Improved offline trained interpretation algorithm
  • New feature: Workspace classifier
  • New and improved OCR service

2021-09-06

  • Finalize improvements
  • Added new field type BIC
  • Updated formatting and validation on IBAN and monetary fields
  • Updated collection templates
  • Minor bug fixes
  • Improve robustness of OCR service
  • Better logging for document with bad text layer
  • New interpretation algorithm specialized for scanned documents (receipts etc)
  • More robust interpretation
  • Improved online trained interpretation algorithm
  • New collection template
  • Updated API documenation
  • Minor bug fixes

2021-06-30

  • New and better OCR service for non PDF documents and documents with bad text layer
  • Finalize item, easier and better management of training of items
  • New collection template ’type-a’
  • UX updates for warning messages
  • Minor bug fixes

2021-06-17

  • Interpreter improvements regarding BG and PG (Less false positives)
  • Duplicate item warning. Status message will contain a warning message if invoice number and/or OCR number exist in another document (for same supplier)
  • Redo OCR correctly on PDF's with a specific type of bad text layer
  • API Documentation updated
  • Improved management of large documents

2021-06-09

  • Offline trained interpretation algorithms improvements (less false positives)

  • Extraction of large documents speed optimization

  • Automatic addition of multiple marked VAT amounts

2021-06-07

  • Minor bug fixes

  • Updated swagger documentation

  • Accounting improvements

2021-05-26

  • New feature: Field status and warning messages

  • New feature: Dynamic document classifications

  • New feature: Finalize endpoint for easier integration

  • New collection template

2021-05-12

  • User activation fix

  • Accounting improvements (improved accuracy)

  • Online trained interpreter improvements (improved accuracy)

  • Standard accounts for receipts

  • General speed optimizations

  • Field matcher improvements

2021-05-05

  • Improved formatting rules for invoice numbers and monetary fields

  • Speed improvements regarding workflows

2021-04-29

  • Improved offline trained interpreter (better accuracy)

  • Currency field improvements (less false positives)

  • General speed improvements

2021-04-21

  • Improved offline interpreter algorithms (better accuracy and less false positives)

  • Improved document classifier

  • Better interpretation of VAT zero and SEK

  • Improved accounting prediction

2021-04-14

  • Fix regarding not writing over interpreted Currency and zero VAT value with previous value

  • Big improvements on predicting/setting debit and credit for accounting

  • Improvements regarding account prediction

  • Extractor speed improvements

  • Online model improvements

  • Speed improvements regarding uploading and interpreting documents

  • General speed improvements

2021-03-16

  • Account prediction: Optimized speed and accuracy

  • Performance improvements: Slight optimization of interpretation time, allround and in particular for OCR

2020-03-10

  • Field formatting improvements (less false positives)

  • Updated offline interpretation classifier improvements (improved accuracy)

  • Faster interpretation

  • Improvements regarding avoiding setting VAT to zero

2021-03-03

  • Increased stability: Internal updates to make sure the service can better handle high loads (no change to the 20-document limit at the moment)

  • Account predictor upgrade: Brand new algorithm put in place which promises higher accuracy on account prediction

  • Opened up api for predefined accounts: GET and PUT for predefined accounts, which are collection-specific standard accounts for total, VAT, and rounding.

2021-02-09

  • New interpreter help algorithm based on statistical methods (more valid documents and higher accuracy)

  • Stability improvements (more robust on managing internal and external services going down or timeout

2021-02-02

  • Stability fixes regarding uploading items (less items that could hang in process state)

  • Retry with longer delays if storage services are not responding (avoid error state for shorter disturbances with external services)

  • Minimizes setting values with just a part of the value (reduces false positives and increased accuracy)

  • Manage more monetary formats (more valid items)

  • Manage more currency formats (more valid items)

  • Manage more date formats (more valid items)

  • Stricter validation of invoice number (less false positives and increased accuracy)

  • Minimizing setting false common values (helps minimizing setting VAT to "0" etc.)

  • Speed up of OCR and mange large documents better

  • Swagger documentation updated

  • Improvements of online trained interpreter (increased accuracy and more valid items)

2021-01-27

  • Stricter rules for monetary fields

  • Accept more monetary formats (more valid documents)

  • Accept more date formats (more valid documents)

  • Interpreter speed improvments on larger documents

  • Accept more currency formats (more valid documents)

  • Improvements on our online and offline trained model (increased accuracy)

  • Improvements on our help algoritms to lower false positives