July 2021


AI Model Improvements

Text Extraction Engines

Extraction engines for the AI Text Extractor are now available on demand!

Your Hero_Flow install process and environment should fit your needs and take up fewer resources.

You now can select the engine or engines that are right your needs to save resources. Once an engine has been downloaded, that engine is stored in your Hero_Flow environment. This gives you immediate access when you need it for your next text extraction model. 

Extraction engine status:

  • When the engine is downloading, the extraction engine status is: WAITING.
  • When the engine is available, the extraction engine status is: ACTIVE.


More Control for OCR and Text Extraction Preview Data

Experiencing inconvenient lag time with an OCR or Text Extractor model?

Both OCR and the Text Extractor have AI model builders that allow you to see a preview of your results, ensuring that you are correctly creating your models.

Depending on the amount of fields you are adding in your OCR model or the amount of questions you are asking your Text Extractor model, you might have experienced an inconvenient lag while Hero_Flow processes those results.

Use the Preview button to reduce that lag time by toggling off the preview data until multiple requests have been made. 

 

OCR Studio: View, Edit, Remove, and Filter Template Fields

The OCR Studio has been improved with an updated inspector panel. 

You can now enter filter strings to find fields, filter by field page numbers, and remove fields in bulk.

Removal of Flow Studio v1

Hero_Flow's original Flow Studio (known as Flow Studio v1) has been removed from the application. Long live Flow Studio v2.

We're Retiring AI Studio

The Hero Platform allows automation developers/designers or citizen data scientists to create and tune AI models that can classify and extract information. 

As we improve features and capabilities, you'll no longer see the AI Studio within the application. We've received overwhelmingly positive feedback for OCR Studio, Classification Wizard, and Extraction Studio – all vital parts within the Hero Platform to support better understanding of the widest range of documents! 

Custom AI models are still supported by Containerized Functions, and all users will get the most current functionality available.

New Functions for Greater Flexibility

Delete Tuple

The Delete Tuple function capability allows automation designers more control on what point the tuple records should be deleted from the Tuple Store when processing responses from a Robin Input. The function increases flexibility to handle interactions with Tuple Store data anywhere it is used in a Flow.

This function gives you more control over deleting data from the Tuple Store than the previous method in the RobinSkill Input and TupleStore Input. The configuration field "Post processing action" has been deprecated from the Input configuration. 

SplitPDF

The SplitPDF function splits a multi-page PDF (binary) into a list of PDFs (list[binary]) consisting of all of the individual pages of the original PDF.

JsonParser is now Parser

We have updated a parser function (JsonParser) giving it additional parsing capabilities. The new function is called Parser.

Email Output Variables for Rich HTML

The syntax for adding variables in email Outputs has been updated. You're in control of the decision to inject your output variables as HTML or as plain text!

  • ${fieldName} syntax for injecting HTML parts.
  • {{fieldName}} syntax for injecting text parts. (In text parts <,>... characters will be rendered as text, not as HTML tags) 

Prompt to Name Automatic Outputs

When creating a Flow in the Flow Studio, you have an option to use Hero_Flow's Automatic Output to save the Flow data in the Tuple Store. Instead of having to rename your Automatic Output after the Flow has been completed, you can simply name it right as you create it.

Drag-And-Drop Binary Files in Robin

We have made it even more convenient to upload files in the Robin application. You can now drag-and-drop attachment files directly from your computer into a Robin task details page.