Adding High-code to low-code solutions
Rabobank uses PowerApps to modernize and automate certain business processes. Developers usually develop their PowerApps application in combination with some Power Automate. But did you know you can actually run an entire .NET-stack ‘inside’ PowerApps?
Imagine having not to use only the connectors provided to you by flows, but be able to communicate with everything. With just a few simple steps you can link your low-code application to any resource you like and perform every operation you can imagine. Use machine learning options in the Power-FX editor like you would use User().Email?
I got you.
The basics of communication
PowerApps and Power Automate rely on REST API as ways of communicating with external resources. To make connections with external data sources in a safe manner, a Power automate flow normally triggers another flow via HTTP. So your connections and credentials are implicitly shared. Removing nested flows increases safety and performance. One way to achieve that is to use Azure Functions:

Creating an Azure Function
In Azure, you can create a function via the Portal. The function can run in C#, PowerShell or even Java. Choose the HTTP request template, fill in the required information and your demo is already up and running. You can then use it to call your machine learning application, do encryption or even manipulate data strings.
Creating a custom connector
To communicate with Azure Function, a custom connector needs to be created inside of the PowerApps studio. Create a new custom connector and fill in the following basics:
- The host URL, which can be seen in your overview in Azure.
- If applicable, your API key
In the next steps, you can create the operations you need based on your input. Looking back at the User() operations, User() can be seen as the connector, where .Email, .Fullname et cetera are the operations specified. After specifying all operations you need, test it and save your connector.
Use your operations in a native experience
Now that the custom connector is working, let’s use it in an app. Create a canvas app and add the custom connector in the data pane. On a button, for example, your new function can be appended and used like the functions you’re already familiar with!
Moving forward from here
Now that you can create and implement Azure Functions, there’s no limit to what can be achieved. Machine learning, natural language, block chain; it all can be reached in a platform that was very limited for most users at first. So the next time someone says: ‘it’s just not possible!’, you know you can make it work 😉