Add-on services

Robot thinking

Adding another layer of innovation

What are add-on services

Some web APIs charge a fee per use, usually based on the size of submitted files. For us to include these in Tag, we need to calculate fees on a job-specific basis. This is done using add-on services.

Our customer billing portal allows you to purchase subscriptions for Tag. It also (soon) will let you purchase add-on service credits. After purchasing $10 worth of credits, you can consume add-on services up to that amount. In the future billing may be handled differently.

What add-on services are available

The focus of Tag's add-on services is providing access to Machine Learning models. These are services where you submit one or more files and receive back some kind of value-added result.

The services listed below use world-class machine learning models that are available to use in pipelines.

  • Optical character recognition
  • Entity detection within formatted or plain text
  • Text analysis to detect sentiment (the mood of a document), key phrases, language and syntax (e.g., find nouns and their related verbs)
  • Translation between a wide range of languages
  • Topic modeling – scan a collection of documents and establish common themes or subjects
  • Text-to-speech using the same technology as Amazon Alexa
  • Speech-to-text which can be used to transcribe call center calls or other voice recordings into text

Add-on services are consumed using custom steps in a pipeline. When you provide input files and run the pipeline, a cost for that job is calculated and you are prompted to confirm before proceeding. The results of the call can be handled like any other content in a pipeline, and processed further as needed.

Optical character recognition example

Optical character recognition refers to the process of scanning an image or PDF file and extracting the contents as text.

The machine learning model behind this service is sophisticated and works well even if image quality is poor, or if the image is rotated as shown to the left.

The scanned text that is returned can be inserted into generated content, or otherwise processed by pipeline instructions.

Entity detection is used to detect people, places and commercial items within emails, documents or any other source of text. Detected entities are categorized, scored and located in source text.

When using the returned results to create generated content templates, entities are selected in a grid to highlight the term in text. You can then replace each term with a data value or logic instruction.

Detect entities in Scribe
Add-on services and onboarding

Tag's support for add-on services is quite new and will be improved further over time. We are interested in building no-code tools that makes working with these and other advanced services easy and intuitive.

Let us know if you have any suggestions or requests related to this area. We also have tips and tricks that we can share during onboarding which can help you get started quickly.

Add-on services are available to Community and all other subscriptions on a pay-per-use basis.

nSymbol logo
SUBSCRIBE TO OUR NEWSLETTER