Testing your learning instances before moving them to production ensures that such learning instances meet your expectations in terms of the performance, quality, and reliability of the extracted data.

Starting from v.36 release, as a Document Automation developer, you can use the Test mode option in learning instances to try different settings, such as using different OCR providers, 생성형 AI providers, different 생성형 AI prompts, and turning validation feedback on or off.

Note:
  • Test mode is recommended for use only on learning instances in non-production environments such as test, development, and UAT.
  • Test mode is disabled when you export or import learning instances that have test mode enabled.
  • 클라우드 추출 서비스 is not supported in test mode.

Benefits

Track all changes using versions and extraction results in a single console
A single console for developers to track all changes as versions for learning instances and assess their impact of changes on extraction results.
Compare results across versions
Developers can compare extraction results across versions and can compare output after each change. For example, comparing extraction results with 생성형 AI option enabled or disabled. This way developers can choose the most optimal learning instance settings.
Quicker testing cycles
30% faster development experience due to optimizations that significantly reduce time to process each page for developers. This enable quicker development cycles by reducing the time required for testing significantly.

End-to-end workflow

The following image shows the workflow of how a learning instance is tested in the Dev environment before moving it to the UAT environment:

End-to-end workflow of testing learning instances in the Dev environment
  1. Create or edit an existing learning instance. See Document Automation에서 학습 인스턴스 생성.
  2. Enable test mode for the learning instance. See 학습 인스턴스에 대한 테스트 모드 활성화.
  3. Configure the learning instance and save the changes. See Document Automation에서 학습 인스턴스 생성.

    A new version is created whenever you make changes to the learning instance configuration.

  4. Process sample documents to validate data extraction results. See 테스트 모드에서 문서 처리.
  5. Validate the processed document to check if the data is extracted as expected. See 테스트 모드에서 문서 유효성 검사.
  6. Compare the data extracted from different versions of the learning instance to find any configuration changes that might improve data extraction. See 학습 인스턴스의 버전 비교.
  7. Repeat step 3 through step 6 until you get the desired results.
  8. After you get the desired results, export the learning instance from the development (Dev) environment and import it into the UAT environment. See 학습 인스턴스 내보내기 및 가져오기.

    Perform an end-to-end testing of the learning instance in the UAT and Prod environments before moving it to production. See 학습 인스턴스를 프로덕션으로 게시.

    Note:
    • Version history is not imported when you export and import learning instances from one environment to another.
    • If you are using validation feedback and if you want to use this feedback for testing, ensure that you synchronize (sync) the learning instance from Prod to UAT and Dev environments before starting new development.