Hello and welcome to my podcast Azure Tech Talk. Last month I attended Microsoft Build conference held in Seattle. Obviously I didn't go there physically but it was on online. Their experience was amazing and this is one of the conference which I always looking forward for it. You learn a lot of new things, technology, Microsoft releasing everyday and you probably hear about like what's new coming out in, in this kind of conference.
So I'm, I'm not going to go through all the texts which they have released and announced where in, in that conference. But what I'm going to do is like looking at my some of the favorite things which I wanted to share with you, which which I think is brilliant probably going forward because what's happening currently right now is everybody is going towards
artificial intelligence. So you look at Amazon, you look at Google, and then you come back to Microsoft is building huge things in using the artificial intelligence. In my personal opinion, what what's happening is now is they are making AI accessible to everybody rather than it's becomes difficult to utilize it and incorporate into your application. It's now becoming easier to incorporate in your application, making it more artificial intelligence so it can make decisions for you.
I'll give you a very quickest example with Azure AI Foundry. It's it's now very easy to build those kind of models and train them with your custom data. And on top of it, it's it's easy to build agents using your those models now. And it's now in your fingertips. You don't need that much kind of knowledge. It's all in user interface and with very basic level how to build the model like kind of user guides are available for you now.
So once you build those models, they provide you REST APIs which with easy integration into your existing application, or maybe you want to build something new, you can incorporate that kind of model into your application. And that model is hosted in Azure. You only pay for the compute process, the model it takes to actually return you the results
or do the computation power. But what I'm telling you here is like back in the old days where you have to be a core Python developer, need to have an understanding of like how the AI actually works. Then you build the model, you tune it and then you train it with your data. Now this is become really easy for everybody to just go on a journey of learning artificial intelligence and start building, utilizing ready to meet ready made models.
Just train them and use them in your application, which is, which is incredible to be honest, like developers. Now you talk about the developers who are front end, back end developers or full stack developers wouldn't take that much time to actually go and learn on a basic foundational AI course, learn about AI Foundry and learn about the basic concept of how to build and how to actually consume a bottle. So it's, it's more about like
consuming rather than building. So you train the model with your data and it's ready to be utilized. So anyways, like few things to discuss with you because there was plenty of things they discussed explored in their conference because it lasted for two to three days. But the key things which actually I'm going to share with you is about how they have evolved the GitHub copilot, according agent. Now why you can do that. The GitHub copilot can do a range of development tasks for
you autonomously. It can refactor your code, improve your test coverage, fixing defects and even implementing a new. It's becoming insane. For example, a copilot agent can ultimately help you do the stuff. So it means like from the productivity point of view, you will be able to deliver your work faster than ever before. And therefore for even complex tasks get a copilot will be able to collaborate with other agent across all the stages of software life cycle.
So what they are trying to do that this software development life cycle, it's being incorporated by this new artificial agents which can help you do the work quicker. So I'll give you one example. Like when you develop a piece of software, obviously you need to write some kind of unit testing and basic integration testing. Rather you spend that time writing that integration or unit testing, the copilot can help
you out. And in the future, like if you're going forward, like some, some bugs appeared in your application in that specific area, what it can help you out is to fix that bug automatically and even create a PR for you. So I think this is this is one of the biggest step from the get up copilot, which which ultimately is going to help the developers to speed up their delivery. And if you look from the bigger picture point of view, a customer wants some feature to be delivered in X amount of
time. Now it's possible because it can help you write with code, but then where the human intervention comes in is to validate like the code which has been written is actually conforming to the customer requirements as well. So that was one of the one of my favorite kind of announcement from the Microsoft Build
conference. The second one which is about like modernization capabilities in GitHub Copilot means like in the past, if it used to take you a couple of days, months or weeks probably to migrate your existing application to the new kind of framework forexample.net Framework 3.0 to dot net core 8 or 9. Because you have to touch all the code base, look at the libraries, the packages which you are using, how are you going to transform them into the new
kind of framework? It was the kind of time consuming like with the new announcement of GitHub Copilot mode app modernization can it can do all things for you, even your Java apps, your.net apps able to organize using the faster UI agent and it can do the coolest assessment and even remediate any kind of problems happened during the migration from old framework to the new framework, even configuring your dependencies across hundreds of thousands of files.
So means like the work which used to take you about like a month can be done in a week or maybe less than a week. I've not tried it personally, but it's in public beta so you probably going to go and give it a try. And I've got few applications with me but which has been written quite a while ago. I'm going to give it a go but but this is amazing. I think. I think when when people think about modernization or transformation of their existing application, my personal kind of
advice would be forced. You need to make sure like your application has covered a substantial amount of unit testing and integration test because it will help you validate your customers requirements. And then once it's done that then the next step is like how you can upgrade that application to the latest framework. Not like I'm talking about from the architecture point of view, but as it is how it can move from the older framework to the
newer framework. I think this is the kind of work where the developers feel not very happy because it's a very kind of Labor work. You have to go through each file dependency libraries and make sure that you can you can actually manage them easily. But with the modernization tool, you don't need to worry about the copilot will do all the things for you. The thing where you have to come and help it is the manual intervention to make sure that things are working exactly like the way.
The last thing about this one is the copilot tuning in the Microsoft 365. I think this is a very good and it's it's a game changer probably. So I can do that. You can you can use your knowledge here to train the models on in on your company's data on a specific domain. For example, either it's accounting, financial or HR or maybe onboarding a new technology, or maybe you're buying the stuff from the market and you have like a legal and they can feed in all the
recommends to them. And then what happened? That agent will help you make the decisions right? It can help you speed up the work for you. So for example, you're purchasing a software from a third party company and you have a set of procedures rather than you fine tune those procedures or write those documents to actually on board those that third party kind of software
where you could do that. You can feed it into this Copilot agent with an the 360 app 5 app, which can help you actually generate those kind of documents and even help you write those kind of business cases to actually purchase that third party software. I think this is huge in terms of like helping the colleagues, helping the staff today, day-to-day staff, helping with their day-to-day kind of work.
I think it's, it's, it's, it's a big thing, big step forward and it can definitely, definitely going to help and the staff and the colleagues within, within an office environment to help and speed up the process rather than writing everything from the scratch and following the process. This is where the tuning can help you out and incorporate into your existing processes and it can help you even with your
Q&A, all these kind of thing. And you don't need to worry about like looking into something which is probably in in your old documents to find answers. Once you feed in all the information to that copilot tuning, it can help you and retreat those information easily, even do the work for you. So I think these announcements are really great.
I'm really looking forward to working, but then then exploring further And I thought like, maybe it's it's, it's good to share with you if you haven't watched it or if you have already watched it and you're trying to explore something. But these are my favorite ones, which I think is great. AI is changing, is changing the world. We have to embrace it. We don't need to be scared of it, to be honest. The world is changing. It's a kind of evolution happening.
But rather than being scared, we embrace it. We learn about these things. And I don't personally think like AI is going to replace your, your, your kind of job. But what I think it's like, it's definitely going to speed up because when you are delivering A requirement to your end user or your customer, the most important thing for them is like how quickly that can be delivered to us, right? And you always think, think about giving the quality and the value back to your customers, right?
And if you don't give that value in a very specific time, which your customer has asked for, so you are upsetting them and it means like they might lose a trust on you. So probably with these AI tools coming up every day, enhancing your work, I think it's it's helping you to speed up that process, speed up the delivery of your end user kind of requirement, but it also gives you the flexibility to actually help and improve that process manually verifying from the
human point of view. Overall, I believe like these are going in a very positive direction and I'm continue to learn more about like how it's going to shape this world. I think that's it from me on this one. Do shame here with your feedback. How did you find out and do do let me know like if you are exploring about AI or anything, anything exciting, I'll catch you up in my next podcast. Till then, thank you. See ya.