AI Builder - Take full advantage of the prebuilt models

[ZePowerDiver]

AI Builder

It's been a few months now that AI Builder has been released in preview within the Power Platform and this month, some of the features it contains are released in GA.

Besides the Business Card Reader component, all of the presented options require training with historical data, images or documents. This allows the models, such as Prediction, Text Classification, Object Detection and Form Processing to be adapted to your organization's specificities. But it also means that you have to gather that data, understand it, classify it and process it within the AI Builder tools.

Custom AI Builder models

What prebuilt models bring to the table are options to answer use cases where you can leverage common knowledge and expertise that Microsoft has already cumulated. The early example that was already available is the Business Card Reader component in the Canvas App maker.

Business Card Reader Component

With this component, you have the ability to take a picture of a business card in your Canvas App and get the extracted name, email, phone number and other valuable data to push it over within your CDS leads for example.

New Prebuilt Models

Now in preview, a new series of prebuilt models are available as well as extensibility of the Business Card Reader. Here's an overview of what these are about and what they're used for:

• Business Card Reader: allows you to extract contact information from business cards images. ** Now available in Microsoft Flow and Model-Driven Apps (including Dynamics)
• Key Phrase Extraction: identifies the main points in a text document, it can extract a list of key phrases from unstructured text documents.
• Language Detection: identifies the predominant language of a text document.
• Text Recognition: can be used to extract recognized words from documents and images into machine-readable character streams.
• Sentiment Analysis: can be used to detect positive or negative sentiment in social media, customer reviews or any text data you want to analyze.

List of Prebuilt Models

How to use?

Currently, besides the Business Card Reader, all these prebuilt models are only available for integration in Microsoft Flow.

If you want to use them, you will need to be in an environment located in a region that supports AI Builder and create your Flow from within a solution.

Create a Flow within a Solution

Then in your newly created Flow, select the Predict action.

Predict Action

Select the model you want and provide the parameters based on the model (see above for official documentation with details).

Sample Predict Action for a prebuilt model

Next step will be to act upon the results. For this, you will need to use the JSON Parser, once again using the documentation as provided above.

JSON Parser example

Once results are parsed you can act upon the obtained data. GO and experiment!

Don't hesitate to comment and provide feedback.

More information

Here's a quick overview of the different AI Builder models and components with indications of what can be used in which area. For more details, visit my blogpageAI Builder overview

About me

I'm Éric Sauvé, aka ZePowerDiver, solution architect and team lead at XRM Vision. I'm a community leader for TDGi's global hackatons in Canada. Power Platform enthusiast, delivering App in a Day sessions, leading local community events such as Montreal Power Platform Happy Hour, actively participating in Montreal Business Application User Group and speaking in Power Platform and Dynamics CE related conferences.

You can reach me on:

LinkedIn: https://www.linkedin.com/in/zepowerdiver

Twitter: https://twitter.com/ZePowerDiver

Facebook: https://www.facebook.com/ZePowerDiver/

YouTube: https://www.youtube.com/channel/UCBEIW1Pzg2dAJD1YukaE2pw

Blogger: https://zepowerdiver.com/