This June, Microsoft announced its AI Builder at the Microsoft Business Application Summit. AI Builder is a low-code artificial intelligence toolset that is directly available and integrated into PowerApps and MS Flow.
This release of AI functionality by Microsoft gives the power of artificial intelligence to Power Platform users - from pro developers to frontline workers.
In this post, we will cover top solutions for implementing AI in Microsoft Flow and PowerApps. Some of the highlighted solutions leverage the new AI Builder functionalities - others leverage Azure AI services.
Power Platform AI Builder
In case you aren’t familiar with the AI Builder features, here is a rundown of the Five main features launched in the initial release:
Binary classification: This the platform uses historical data patterns to classify your new data into two categories. It’s great for predicting Yes/No business outcomes.
Text classification: This feature uses historical data patterns of previously labeled text items to tag any new snippet of text for your business-specific categories. This helps in streamlining businesses by adding relevant tags to new text.
Object detection: This feature works with images to count, locate and identify objects inside images. This feature can be used to automate different kinds of tasks in various industries.
Business card reader: This feature is available in PowerApps to scan business cards and extract contact information from it.
Form processing: This helps extract information from forms, as per a matching sample form you provide. For example, it helps in reading tax invoices.
The Blogs and Videos
1. Advanced | Flow of the Week: Automating Intelligent Customer Service using Microsoft Flow, LUIS.ai and Dynamics 365
Author: Kent Weare
In this article, Kent shows how Microsoft Flow, LUIS.ai and Dynamics 365 can be used systematically to service an increasing base of customers without increasing head count or labor costs
The tools used help capture the intent of the customers and identify the priority or urgency of the issue described in the emails received from customers. Identifying intent in customer communications helps in routing the issues or queries to the appropriate department or person to solve issues in a timely manner.
The flow of the platforms includes:
Filtering emails for specific keywords in MS Flow
Interpreting the intent of each email through LUIS (Language Understanding Intelligent Service)
Logging each issue as a Dynamics 365 Case (as per the identified intent), to inform the concerned department
Sending acknowledgement emails to the customers with their case numbers.
Author: Jon Levesque
In this video, Jon and Kent build an app called ‘Hot Dog, Not Hot Dog’ using Microsoft Flow, The Custom Vision API, Flow Buttons, and SharePoint.
In the video, you’ll see how to train a Computer Vision model, which feeds the artificial intelligence of the system to identify whether an image contains a picture of a Hot Dog or not.
Author: Leila Etaati
This blog post shows how to create a face recognition app with Microsoft Flow, Power Apps, and Cognitive Services Face API. This is the last part of a four-part blog series.
This is part of a four post series that covers how to make an app that includes the following features:
Capture Pictures in PowerApps
Submit pictures captured within PowerApps to Flow
Use the Cognitive Services face recognition API
Connect Flow to PowerApps
Display the results of the AI
Author: Michael Blythe
This post covers step-by-step development of an App with Cognitive Services Text Analytics API. Here is a runthrough of the post:
Starts with an introduction to the Text Analytics API in Azure.
Shows how to connect the Text Analytics API to Text Analytics Connector.
Shows how to call the Text Analyitics API from within PowerApps to detect sentiment.
Author: Thanura Wijesiriwardena
Microsoft Cognitive Services offer an AI tool called Computer Vision. Computer Vision’s main features include analysis and description of an image including Object Detection (with bounding boxes), classification / tagging and description with probability.
As opposed to the Custom Vision API mentioned in the Hot Dog / Not a Hot Dog example above. The Computer Vision API is not trainable by end users.
The blog shows how Dynamics 365 can be used step by step to read data in receipts, and how to label/tag or describe images uploaded to Dynamics 365.
Author: Bruno Capuano
This is one of the most simplest yet exciting blogs. This blog shows how you can configure MS Flow and Computer Vision Cognitive Services In just three simple steps to create a mobile app to analyze and describe images.
Author: April Dunnam
Here is a video that demonstrates the AI Builder Object Detection Functionality - it shows how you can use the Common Data Service with AI Builder’s Object Detection feature in PowerApps.
Upon creating an AI model, select the entity you created in Common Data Service, add and tag images, and train the model. These steps will give you a model, to which you can upload images and get a response in terms of detection of the object you created the model for.
Author: Vivek Bavishi
In this video, Vivek Bavishi explores AI Builder’s features including scanning a business card, processing an invoice, and object detection.
He shows a demo of these three capabilities of the AI Builder a PowerApp. Through this overview, you can visualise the end results that can be achieved with these AI features.
Author: Reed Fansler
This fantastic post by Reed Fansler shows, step by step, how to integrate the SharePoint Image Library with the Computer Vision API Cognitive Service that reads images and returns descriptions of the images, their color schemes, tags of images, and other such values.
You will find step by step instructions on how to set up a Flow that acts as a medium to connect a SharePoint image library to the Computer Vision API to return searchable image tags and descriptions.
Author: Vivek Bavishi
Here's another great video by Vivelk that shows how to make a business card reading App from scratch, using AI Builder. After this, he shows how the data retrieved from scanning a business card, using the AI Builder tool, can be recorded in data sources like a SharePoint Library, Outlook Contacts, and CDS.
We thought these posts did a great job of highlighting the strengths and capabilities offered by integrating AI services with Microsoft PowerApps and MS Flow. We hope you find the post valuable for your future AI or Power Platform solutions.