Every month Lennert, one of our podcast hosts, uses Power Platform to build another interesting application. This time he used his creativity to whip up a team building app. The app "The Hunt" allows someone in a team to give other team members hints about the location of a hidden treasure. Each day they can see a new radius that the organizer drew in the app, from which the location can be guessed more and more accurately each time. Lennert used the standard map component for this, which can also display a satellite view of the site. When someone manages to locate the treasure, they can let the others know by using the "I found it" button to share a selfie with the prize. A truly fun and "Covid-proof" team activity!
Work smarter and leave boring tasks to computers
Artificial intelligence can be described as smart software that uses intelligence to accomplish certain tasks normally performed by humans. It is a way to automate boring or repetitive tasks, with results that are faster and often more accurate. This software has become much more amenable and accessible in recent years. It's everywhere, such as personalized content on Spotify or face recognition to unlock your phone.
You can also harness this intelligence on the Microsoft Power Platform with AI Builder. A very accessible way to make your apps smarter. If you want to take things a step further than the available components, you can use Azure Cognitive Services. With these, you can implement high-quality AI models using APIs.
Using AI Builder may seem rather pricey, considering you pay more than 400 euros per month per unit, but the cost often fades once you calculate the ROI. Building in AI allows employees to get rid of repetitive tasks and frees up time for tasks with more added value. If you are looking to get started with this, you should first check whether your use case fits in the per-app or per-user plan. If you are thinking bigger and want to buy units per month, you should use the AI calculator to determine how many you should buy. This tool helps you estimate how many units you will need per component of AI Builder, for example based on the number of invoices you process per month.
Our experts' favorite AI Builder components
Form processing: processing data
You could use this component, for instance, to automatically recognize data on a photo or scan. One example would be to let employees scan in sick notices. In most companies today, this is often done by sending an email with an attachment to the relevant department, which then has to manually record the employee's sick leave somewhere.
AI Builder recognizes the data entered on such a notice, allowing you to transfer the relevant information to the necessary systems or automatically set an out-of-office message including the length of illness. This model can be trained with just 15 sample photos, but of course the rule when training an AI mode is the more, the better.
Object detection: recognizing objects
This involves recognizing objects in photos, such as logos. It can also to count objects, useful when taking inventory. Another scenario where this can be useful is on a production line where different parts come together on a belt. By taking a picture of the belt every few seconds and having AI look at it, you can verify that each part is in fact present.
Category classification: categorizing text based on meaning
This component is perfect for analyzing feedback you receive through various channels. For instance, hotel reviews can be divided into different categories, such as feedback on the location, the food, the level of comfort, etc. You can even add an extra layer and let it assess sentiments, so you instantly know whether the feedback is positive or negative.
What to watch out for: prediction model
This model still has room for growth, but there is plenty of potential. When data is entered, the model guesses whether or not something will happen. It makes predictions based on historical data. Of course, this needs a lot of training and high quality data in order for it to be reliable.
Trained models, such as a business card reader
AI Builder also includes trained models that immediately do what they need to do without you having to train them – the ultimate in low code. For example, one model reads business cards to instantly create new contacts from them. This model could also trigger other actions in an app, such as creating a lead.