nz365guy

View Original

Learning the Power Platform – AI Builder

Microsoft Business Application Summit 2019 announced another build tool – AI Builder.

As many of you know, the world of Microsoft Business Applications has changed. Those that have been in the industry for some time will remember talking about Dynamics CRM, and then Dynamics 365. Although Dynamics 365 was the starting point, today it is just a PowerApp operating on the Power Platform.

For those that used to build applications on Dynamics 365, your starting point for all new application development in our space has pivoted. This change is here to stay.

When building Microsoft Business Applications, we should start with the Common Data Service (CDS) and the three building blocks of PowerApps, Flow, and Power BI.

My approach to career development is to be a ‘learn-it-all’ (rather than a ‘know-it-all’) and this post is the start of my journey of learning the Power Platform, specifically the AI Builder tool.

AI Builder enables non-data scientists to build Artificial Intelligence (AI) enabled applications. As many of you will have seen, AI is on the rise and will replace many jobs that exist today.

The beauty of this new technology though, is that it also creates many new jobs. AI Builder could be the starting point for your pivot to learning and building a career in Artificial Intelligence.

In the next five years our lives will be entwined with AI more than ever. Companies that want to grow and survive in the future will need to augment people with AI to take advantage of the unique capabilities that we as humans struggle with – processing information at scale.

Microsoft has now made it possible for people that do not have a programming background to build Machine Learning (ML) algorithms. Start uncovering actionable insights in the data they already have or will create in the future.

The preview version of the AI Builder is available in North America and Europe.

Currently there are four main models that can be used for various business scenarios.

Binary Classification allows AI to look at data and make yes or no decisions on that data.  Gautham Thapar gave an outstanding example of this in his presentation and demo of AI Builder.

In it, he shows an example from manufacturing.  When an item is manufactured, there are certain tolerances that the product being produced must sit within to pass quality standards. I personally got some insight into this many years ago.

Before I got into Microsoft Business Applications, I went to a little country in Africa called Eritrea to a manufacturing facility that manufactures Intraocular lenses to cure cataracts or myopia blindness. These lenses had to be checked for defects before they could be packaged and shipped to hospitals for use. All defects were recorded on paper, and if there had been an increase in defects, the processes were checked for the cause.

Nowadays you could take these measurements of defects, load them to CDS and in real-time (or batch by batch) understand where defects where entering the process and correct them. The trends and patterns would become very clear as the defects were either a YES over a micron (measure size) level for imperfections or NO under a micron level for imperfections. Causation can be tracked and solved very quickly.

Another example that comes to mind is tracking applications to university to drop-out rates. I saw this study done years ago in New Zealand. If you took the data from student application forms and run it through Binary Classification, I believe we could we uncover the likelihood that a student would drop out of the course based on this data.

Form processing allows digital form data to be analyzed by AI. It can be done with data captured by a camera.

For example, photographing Business Cards and auto extracting that data into a Contact record. Another example is extracting data from PDF form documents, say for invoices or application forms.

When I think about this, there are many possibilities. With Forms ProFlow, and AI Builder, you have a winning combination to process almost any type of form capture that a business may have.

Every business will have a unique application of this. In water utilities, it may be a request for a new water connection that asks a range of questions to arrange for the connection. Or in healthcare, it could be asking various symptoms type questions that could lead to a more accurate diagnosis.

Object detections are used to identify, count, and locate objects in images.

The examples given are counting the various items in a fridge. Straight away from industries such as supermarkets having a use case for this in better stock management, as well as the various suppliers to those manufacturers.

It could also be used to uncover trends in clothing and hairstyles. Or used as part of an identity solution to validate someone’s face. I could see this used to look at data from satellites or drones and analyze those images for information.

In open cast mining, it could be used to monitor progress. In agriculture, it could be used to identify growth and yields from crops. A local council may use it to identify solar installations or swimming pools by suburb.

Text classification can be used to improve processes such as customer feedback. Adding metadata for codification of knowledge base articles.

For example, adding tags to video content based on the video description. Or take this post as an example – at the end of the post, it could auto-tag the post with keywords based on relevance and density of words in the article. It could be used to route requests based on the content in a customer support scenario.

The more you think about, the more ways you can apply it.

Remember this is currently in Preview, but you need to start learning it today. I expect that Microsoft will continue to add to this toolset with color analysis, voice analysis and many more.

The important thing is, now is the time to start learning this new technology. It is a tool to get started in AI with the ability for developers and data scientists to extend what comes built-in to AI Builder.

Remember that AI Builder is built on Azure Machine Learning and Azure Cognitive Services.

To start learning AI Builder look at the videos below, this will be a great starting point and will show you how to get started today.

AI Builder
AI Builder Microsoft Docs

https://www.youtube.com/embed/KOPPubi1X4s?feature=oembed&enablejsapi=1&wmode=opaque

https://www.youtube.com/embed/SK_w601WV0I?feature=oembed&enablejsapi=1&wmode=opaque

https://www.youtube.com/embed/hh9y9IOA-iw?feature=oembed&enablejsapi=1&wmode=opaque