5. 69. Use AIBuilder formulas from within canvas Power Apps
So in the previous video, we had a look at how we could insert AI Builder components. But wait a minute, there are many more AI Builder models that we can use in PowerApps. Where are they? And the answer is for these, you don’t even need to insert a component.
So let’s insert a blank screen. And I’m going to insert a text input and then a label underneath, which will give me the results. So let’s just make it a lot bigger and centralize it. So all I need to do is to change the text to Aibuilder and then I can select which model I want to use. So you can see I’ve got sentiment, categorize detect language, extract key phrases, and text entities.
So let’s have a look at Analyze sentiment. So AI builder analyze sentiment, open bracket. And then we need the name of the control. So text input one plausible bracket. And you can see that we’ve got sentiment, document scores and sentences. So let’s go for sentiment. Now. Interestingly. Lowercase s. All the rest are capitals.
And again, I’ll just put in a background so we can see where it is. So currently it says neutral. Interesting because I haven’t actually run it. So let’s just change this by playing this. And I’m going to say this is great. You can see positive. This is greatly disappointing. You can see the answer is negative. So even though it’s got the word great in it, still recognizes that now it’s an adjective rather than a noun. Okay, so let’s say this is good and bad. We get negative.
Now, you might think that we would get mixed, but here we get negative. So let’s change this to a mi megusta positive AMI. No, megusta should be negative. Coming up yes for audio. Esto bero megusta. I hate this, but I like it. So that comes up negative as well. Apologies to anyone who does actually speak Spanish. Hell there. So that is the sentiment. Analyze sentiment, name of control, and then dot sentiment. Okay, so let’s change this to the next one.
So this is going to be categorized text. So we’re going to categorize text input one categories and we have type. However, there is a problem in a builder. Categorized text categories is a table. And so it still remains a table when we have dot text after it. So type. So what we’re going to do is concat that again. And then at the end, we’re going to say type and a color with space afterwards. So that then combines everything into the table and just gives me the type afterwards.
So now done this. Let’s play this and I will put something fairly long in. So here we have this program is great and has lots of potential. So it thinks I have to do documentation or this is a powerful tool that help makes quick changes. And as you can see, it’s not actually giving me too many categories. This is the pre boat version, whereas if we go back to the boat version, not the one that we can build, you can see that with more classification details. These two examples which I’ve just taken are capable of many more suggested tags. So it is currently in preview.
I think it probably needs a bit more work to be able to really be useful. So let’s look at the next builder category and that is Dot detect language. So fairly easy this one. And you can see we have a dot language. So the answer English. Again, if I change it so that it is Spanish or Diisto, then we have Eschew V French. So fairly easy to use. Well, the next thing we’re going to have a look at then is the extract key phrases. So again, this will bring a table and because it’s bringing us a table, we need to have the concat around it. So let’s get concat and then we put together the phrase.
So again, if I actually just make that a lot bigger, I’m putting some of the phrases that are included in the demo for elsewhere. So this program is great and has a lot of potential. So it gives lots of potential user Interface program link error results equally if I put in it’s a powerful tool that helps me make quick changes. It says powerful tool and quick changes. And then finally we have the entity. So let’s click on this and go to the very bottom one. Extract text entities. So go to the end and type the dots or entities. And we’ve got type and value and length.
So let’s get the type. So let’s play that. So member is looking for things like location and percentages and other things. So it has correctly identified a date, time, temperature and percentage. Let’s go back into this. And I want not just the type, but I also want the value as well. So I think that’s value and type. So now we can see I just put a bit of formatting to make this look a bit better that we have today, which is a date time. So I’m not saying the 21 December anything, I’m saying today 70 degrees Fahrenheit is the temperature and 20% is the percentage.
So in this video we’ve had a look at five different pre boat models and we didn’t have to insert a component into any of them. We just said AI. Builder, notice. Capital A, capital I, capital B. And then Dot. And then after that we have an eye sentiment, categorized text, detect language, extract key phrases and extract text entities. Some of these return a table and if that’s the case, you may wish to consider using concept and joining them all together into one string.
6. Practice Activity Number 15 – The Solution
So how did you get on with this practice activity? So, first of all, I wanted you to create a power automate floor. So this is going to be a button floor because it’s going to run whenever you want it. So an instant floor. So I’ll just skip all of this and call this Ms learn blog. So I want to connect to a feed. But first of all I need a trigger. So it’s whenever I manually trigger a flaw. So next I want to take an RSS action which is list all of the feed items. So the feed URL, while I gave you that in the practice activity. So there it is. So having done that, we need then to extract the key phrases from each message. So each message you should be thinking loop. So let’s have a look at loop or control.
And we have in control. Apply to each. So apply to each and we’ll have a look at the body. So for each body I want you to do something and that something is in the Aibuuda and it is to extract key phrases. So here we have the key phrases. So the language is going to be English and the text is going to be the feed title. So now I’ve got that I need to turn to add it to a variable. So I need to create a variable. So I’ll create a variable earlier on. So I will add an action. So type in valve for variable. I will initialize a variable. So this could be my key phrases variable and it’s going to be a string. And now I’ve got the key phrases I need to add them in. So I’m going to go add an action.
That action is going to be a second apply to each. So for each and here we go down to extract the phrases from text. So for each result that we get, I want to be able to add in. So I want to be able to append to my string variable called key phrases. I want to get the key phrase from the text. So let’s just run through that again. I manually trigger floor. I initialize a string variable. I list all of the feed items from an RSS feed. Now, because this is now an array, it’s more than one item potentially I will apply to each. So I’ll take the body from it. So I’ll take the actual item and I will extract the key phrases from the feed text. And then that might give me lots of different results, lots of different key phrases.
So for each result I want the key phrase because you can see there’s lots of different items for each result. And then at the end I need to send an email notification. So let’s just go for email, send an email notification. So this is my RSS analysis and here are the results. And I’ll put in my variable key phrases so let’s save this and let’s test it by activating it manually. So it started to run. So let’s have a look at it. And you can see it’s going down taking a few seconds. And you can see after 34 seconds it’s finished the analysis and you can see key phrases like power Platform, Functional Consultant, Associate, Microsoft Certified and all the rest.
So I need to have put a comma and a space after each of these terms to make it really better. But here you can see where each starts and ends. Anyway. So this is how we use the AI Builder results in Power automate. So I think the key thing here is make sure you don’t get too many apply to each, but then just enough if you get too few, then you’ll know, because you’ll get an error message. So let’s now move to PowerApps. And this is a very simple app. So I’ll just start a new Canvas app. We’ll make it a tablet layout. And I’m just going to insert in separate text boxes, entities and sentiment. So let’s go and have some text input. And what I’m going to do next is have a data table. So I’ll just go for the data table here, drag it down. Now we need a data source.
So I need to click on data source to go up to this Power bar. But you can see it’s all in items and what’s going to populate it? Well, it’s going to be AI Builder and you can see we’ve got a few options. We’re going to extract text entities and we need the text input reference so that’s text input one clause, bracket entities. So you can see I’m doing all of this without actually knowing what exactly I’ve got to put in. Now there’s no field at the moment, so I’m going to click on fields, see what we get, edit fields, add field, and you can see we have got length, score, start index, type and value.
So I’ll just put in a couple of these. So now we need some actual text. So I’ve asked you to go to the Microsoft website for an article about the PL 100 exam and let’s just copy some text, replay this, put it in here, and here we get some information. So we have got Microsoft Power platform and Power apps being a key entity, microsoft certification, LinkedIn page alert. Now notice that type not 100% accurate. I’m fairly sure that page and alert aren’t cities, but it’s a good start. And now let’s insert a label and this is going to be the sentiment. So is this positive? Is this negative?
So this is again going to be AI Builder. And here we have analyzed sentiment of text input, one dot and we’ve got sentiment. So you can see the answer in this case, is it’s positive? So let’s have a look at some other paragraphs. So I’ll put in the app maker role and certification. So this much copy that into here, control A, Control V, and we can see that this is mixed sentiment.
So we’re talking about key entities such as Microsoft Power, Microsoft Power Platform, Excel, Microsoft Teams, and so on. So in this video, we’ve had a look at how to get the AI builder into Power Apps and Power Automate. And hopefully you’re getting some really good ideas for your own flaws and apps. In the next section, we’re going to have a look at Microsoft Teams. Please join me there.