PTW3 047: Microsoft's AI Architect Naveen Krishnan on AI Innovations and Use Cases with Donna Mitchell

Thanks for checking in the pivoting to web three podcast. Go to pivotingtweb three podcast.com to download and listen or web three game plan to check out the videos. Thank you. Good morning, good afternoon, good evening. Welcome, welcome, welcome, welcome to pivoting to web three podcast. And today we have Naveen Krishnan. He's an AI architect at Microsoft, everybody with a passion for leveraging artificial intelligence to solve real world problems. His journey in the tech world began over a decade ago and we have the privilege of working on some of the cutting edge projects he has worked on cutting edge in AI and cloud computing.

Donna Mitchell

00:00:43 - 00:01:42

Now overall, through his career he's been driven and desires to make complex technologies associated to everyone and accessible to everyone so they understand it clearly. And at Microsoft he focuses on developing and implementing AIH solutions that not only push the boundaries of what's possible, but also deliver tangible benefits to businesses and individuals alike. I'm not going to keep introducing this gentleman. I'm going to let him continue to tell us what's been going on, how he got to Microsoft and let us know what's going on in the AI space at Microsoft and how they're impacting society across many businesses and industries. So welcome Naveen, to our audience. Say hello and tell us how did you end up at Microsoft because that is exciting and I'm so glad to have you here.

Naveen Krishnan

00:01:42 - 00:02:21

Yeah, sure, definitely. Thank you very much for the detailed intro, Donna and yeah, so my career started like 15-16 years back. I was working on a small startup companies in India and I landed in U. S. In 2015 and then I started working for Infosys as a technology architect and I worked there for ten years. And after, after working there for ten years in India and in us, I, I got into Microsoft. So at Infosys I got myself certified as cloud solution architect and then I worked for several enterprise customers like Morgan Stanley and JP Morgan and Chase, things like that. So which really helped me land in Microsoft.

Naveen Krishnan

00:02:22 - 00:02:55

And yeah, after getting into Microsoft I was, I was a CSA at that time and recently when AI started picking up and I got myself certified as AI architect and I helped customers. Today I help around 30, 40 customers of Microsoft to help them get into AI. So I help them design their solutions and I help working with them and I work with them and then I, we developed together. So yeah, that's what I'm doing with Microsoft for last one and a half years now.

Donna Mitchell

00:02:55 - 00:03:19

So since you're with Microsoft, I've got a lot of questions and I'm sure you've got a lot of answers. But most importantly for our audience, a lot of people are still learning some of the technology and the emerging trends. Tell us what is exactly cloud computing? What is that? How did it come into being? What makes it so different from other computing? Can you help us understand?

Naveen Krishnan

00:03:19 - 00:04:05

Yeah, sure, definitely. So in the past, like for example, let's take our enterprise, be it a fintech or be it a retail or any domain. So they do have all their servers and things set up in their local infrastructure, and there will be a lot of investments going to that, right? They have to buy the hardware and they'll have to plug it in with the network security and whatnot, cooling and other structures. It's expensive. It's very expensive. So for an enterprise to invest in this. So that's when the cloud computing came into picture, which is there are a few cloud providers like Microsoft and other cloud providers. They started investing in this and then they created a lot of data centers, which is fail safe and things like that.

Naveen Krishnan

00:04:05 - 00:05:08

For example, if you are, if you are on your own infrastructure and that goes down for some reason, then that is then recovering it and the downtime is very high and you will have to depend on engineer, give a call to engineer, he will wake up and then he will have to run to the data center, plug it in, and then do things like that. So, which is the tedious process, and there are a lot of downtime and their business is going to be down during that time frame until we, until they fix that. So that with the cloud providers, everything is all centralized. And so they do manage and they take care of your infrastructure. And within cloud computing, there are three or four different things. One is infrastructure as a service where you will have some responsibility of all the things which is running on cloud. And with the Paas offering platform as a service where we take care of mostly 70 to 80 percentage of your effort, and then you just focus on your application. So you develop your application and give it to us and we will take care of running and then maintaining security patches and other stuff which takes all hazards from you.

Naveen Krishnan

00:05:08 - 00:06:00

And then you just focus on your application so that you can build latest things and then, yeah, you can be number one in the market, right? So you focus on that and then we focus on the infrastructure. That's where the cloud computing got into picture. And after a lot of advancement happened, happened after the cloud computing introduction. So the integration with other platforms, like for example, I'm a bank, I wanted to integrate with the retail application. And if they are on cloud and the integration is going to be easy and smooth and they can talk very quickly and then they can talk very securely. So there are few few ways where you can connect through the data centers backbone. You know, even you don't have to travel in the public network, right? So there are a lot of capabilities for, in terms of security or in terms of reliability and things like that. So which brought cloud computing into the, into this.

Donna Mitchell

00:06:01 - 00:06:53

So when you have cloud computing and then you have all the different platforms enters AI, there's an intersection in that space. What happened or how does AI really make an impact today? We see it happening in all different industries, sectors, businesses. There's really a serious transformation. But you have the cloud computing and you have Aihdem. How did it end up coming into being and then how did it just take off? It's like one of the real technologies that came in and it's making such a significant impact and it's, and it's touching everything and everybody. Can you help us? Those that aren't in the development world, in the application world really understand how that took place. It's like where did it come from and how did it get here so fast?

Naveen Krishnan

00:06:54 - 00:08:09

Yeah, so it's not all this happened overnight, right? So it took time like especially with partnership of Microsoft, with OpenAI. So that's when all these things started evolving and then we, Microsoft invested and then open a got into picture. So there were like lot of studies and a lot of papers have been established, published over the long time. But how to run on how to make this it and the growth of AI is very quick when you compare to cloud computing. When cloud started, it was started by Amazon and then it evolved slowly and it took around 810 years now to have people comfortable on cloud. But with the recent introduction of AI, when AI, just like one year back, AI got it and within a year everybody started investing on copilot and everybody started working on AI. So which is very interesting to see people diving deep into AI and then doing lot, many things and even it has got that capability to handle all this requests and things like that. So especially with this partnership, I would say.

Naveen Krishnan

00:08:09 - 00:08:38

But apart from that there are a lot of new releases happening with open a. So when we started it was like it can do only certain number of things, but after it, when it start evolved, within a year it can do so many things and then even I don't know how the next year and two years are going to be. So this is the fastest technology what I have seen in my career.

Donna Mitchell

00:08:39 - 00:09:29

So for someone that's listening or looking at the videos, because this is going to be audio and video. When you say so many things have taken place, so many things that you've seen, and it's been so fast, can you give us or paint us a picture of when there was no AI? And now that same industry or sector of business has AI and what were the differences and the significant impact that you've seen that brought that to the forefront and how Microsoft's genius and skill sets and employees and just their mission and vision and all the power that they bring to the table, how that impacted. Is there a project that you can think of that was quite challenging with AI in all the advancements, it's really made a significant impact.

Naveen Krishnan

00:09:29 - 00:10:14

Yeah, sure. So I can talk about one thing, what I just recently worked on. So there are so many evolvements. I'll just take one small piece so that it can be, so it can reach every, all the level of audience. Right. So that's what I wanted to focus on. So for example, if I run a, I ran a retail shop and then I wanted to see like how many orders of this particular item has been sold. And so for, to get that report generated, I'll have to go to my development team and then I'll have to work with them for like maybe one and off two months to get that report to me, report delivered.

Naveen Krishnan

00:10:14 - 00:11:26

And then, so, but with today's involvement, I can convert my natural language into backend queries and then, which can go grab the data, what I want. So I just type in my natural language, say, like I can give me the list of orders, what I have sold today for this particular product and it's going to go to your database and it's going to collect all the database tables, schemas and things like that. And then it's going to pull and then give it to you. So that's one simple thing. What I worked for one of my customer last week. So every month I do at least three to four innovations with around AI. So most of them are not, not just to one specific phase, but for retail and then it's for manufacturing and it's for things. So we have developed one solve one use case where you can the gifts on the father's day or for, or any of their Christmas gifts, right? So where do you want, where you want to print your, uh, print your own picture and then you want to customize your gift so it can be done quickly and easily where you can now you can go ahead and select.

Naveen Krishnan

00:11:26 - 00:11:48

So these are my parents and these are my this. And can you make some funny faces around that? And then you can customize your product. So those types of things also available. Right. You can create innovative images on top of something. So on top of your customized kit. So that is one, one thing. What is what I have seen and then very impressive.

Naveen Krishnan

00:11:48 - 00:12:16

And apart from this, these are all small use cases, but at very high level, I have seen a lot of enterprises and a lot of manufacturing and retail companies, right. So they are getting benefited with several mini use cases. And so there are like several open forums where people are discussing about these innovations which is happening around these larger industries. So yeah, there are so many use cases which is getting developed every day.

Donna Mitchell

00:12:16 - 00:12:24

So what are some of the, what are some of the bigger use cases? What trends are you starting to see now that's taking place today?

Naveen Krishnan

00:12:25 - 00:12:52

Most of them are around retail is what I see to start with. So that, that's up to my knowledge, actually. So when you see. Right, so when you wanted to. So yesterday I had a problem. So I saw a, I saw a product and then I wanted to. So that's broken in my la, in my lawn. And then I don't know what its name and it's a plastic lid of some, some pipe.

Naveen Krishnan

00:12:52 - 00:13:13

So it's broken and I don't know what product it is and I don't know how to search. So with the new advancement, if you just show your product and then it kind of recognize like what type of product it is and where it is available and how you can buy even, it can create an order for you and then it can, it can, it will be delivered. So these are the advancement what I see in the retail space.

Donna Mitchell

00:13:14 - 00:13:50

So with you working with Microsoft and AI, and you probably work with hand in hand and see so many intersections and use cases. With today's technology, what has really concerned you or raised your eyebrows and you wondered about, is there anything that you found to be something that could be a negative or something that you get concerned about in regards to it being used unwise or in a manner that's inappropriate and could have a negative impact on society?

Naveen Krishnan

00:13:50 - 00:13:57

No, I don't think so. With the recent developments and there are several guardrails that's been put in place.

Donna Mitchell

00:13:57 - 00:14:02

Oh, yeah. What are the, guard, what are some of the guardrails? That'd be nice to know some of the guardrails.

Naveen Krishnan

00:14:02 - 00:14:50

What I can see is it can filter out some sensitive contents and it can take off, take away. You can, for example, you can take few of your what you say. So that depends on the business, right? So if you want to avoid these types of content to these types of audience. So then that can be filtered out and even if you, the level of access, what you give to your data can be controlled. So for example, I am, I work for a company and then I have my data and then I have my copilot sitting and if somebody, some of my employee try to hr database. So that's not what, that's not allowed. So these types of things are restricted by the user level of access. What level of access you have today to see your, to the company's data.

Naveen Krishnan

00:14:50 - 00:15:17

So the same level of access can be granted to your copilot as well. So these types of things and yeah, so that's, that's a very high level. What I can, what I can tell. But there are a lot of documents available even today, right. If you wanted to know about these guardrails, there are a big list where it contains numerous amount of data, what can be and how and at what layer it can be. So these are all available.

Donna Mitchell

00:15:18 - 00:16:01

So with that said, as we go forward in today's technology, there's a lot of conversation about AI and cybersecurity and how does Microsoft work in that cybersecurity space with AI? Tell us what we need to know today that's going on that you can share in regards to some of the projects and there's so many different, you know, hacks or systems going down and companies having issues and just so much happening. How is Microsoft playing in that space? I know they lead in a lot of areas. Can you share some insight on that with us as well?

Naveen Krishnan

00:16:01 - 00:16:51

Yeah. Security as a whole, right? Not just AI, security as a whole is, is a key. So everybody should focus on their infrastructure security. So be it at any level, right? To the level of access, what people has to your data and to where you are open to. So those level of access has to be taken care by the enterprise itself. But apart from that, when your data is on cloud or when your data is on with all the proper restrictions and things like that, it is all completely secured and then it's all say nothing. No data is open. And even as a, those data cannot be seen to the, seen by any external people apart from those who have access to.

Naveen Krishnan

00:16:51 - 00:17:18

So that level of security is already in place apart from the, in terms of AI. So I recently heard that we have, we do have a security copilot, so, which is basically designed for your security which can go identify and then can give you all the data of the level of security, what you have got. So that is also available is what I can see, but I have not explored into the security copilot yet. But I will have to deep dive once to see what it has got.

Donna Mitchell

00:17:18 - 00:18:15

So what separates Microsoft's copilot, what separates the different AI's but Microsoft's AI OpenAI just everything. What makes you always better? What makes you the best all the time? Can you share any insights or is it part of the mission, the culture, the skills, the intellectual genius? Or tell us what really is it that, I mean Microsoft has been around since I don't know when and it stays in the game. I mean what makes that happen? How is it that that always happens? They stay in the game, they stay ahead of the game. And even if it looks like they got kicked to the side for a minute, they come back roared. What makes Microsoft who they are and always seem to be with the winning hand?

Naveen Krishnan

00:18:15 - 00:19:01

Yeah, so it's all, I would say it starts with the vision. So basically how, what type of vision you have? So you might have seen, right. So every organization, every world. So if you read our CEO's vision, like, so it has got a broader vision and so the investments on AI is one of the key thing. What has just got, got a big boom. And so apart from that, so there are like in every sectors basically where technology can evolve better and evolve more. So for example, in gaming especially, so our recent acquisition of activation Blizzard. So that's one of the key thing for gaming.

Naveen Krishnan

00:19:01 - 00:19:44

So where a lot of people spend their time on the gaming. So especially after this Covid right after pandemic. So that's one of the thing. And apart from that a has also got a big boom and so, and also the devices itself. So you are, if you have, if you see the laptops now which comes out, it has got a copilot pre built on the laptop. So those devices as well that, that started creating new era. So, so these types of things. And I, yeah, also the culture, as you said, which all it's all together and then it's coming up as on Microsoft.

Naveen Krishnan

00:19:44 - 00:19:46

So that's what it is.

Donna Mitchell

00:19:47 - 00:20:55

Do you know anything in regards to any projects on the scale that really impact nonprofits and how they handle some of the crisis in the global communities and I, some of the disasters, can you speak to how that would impact or how it should be getting used more frequently on the nonprofit side of things? Can you share any information on that? There's a lot of nonprofits that listen to this podcast and they're starting to really look at how they handled donors and the recipients and blockchain technology, and they're starting to look at AI. What insights can you give not only to entrepreneurs and big brands that work with Microsoft, but some of the national foundations and nonprofits that really spearhead major projects. How are you working with those type of organizations?

Naveen Krishnan

00:20:56 - 00:21:26

Yeah, so one thing, I have one customer whom I'm working with. So that's, that customer is supporting the nonprofit. So he is, he's actually a customer for customer. So he supports a non profit or, and then he does a lot of development. So what I have seen is main focuses on the cost saving is what I can say. They start with that. So I work with them from that perspective. So basically how that happens is when you say a nonprofit.

Naveen Krishnan

00:21:26 - 00:22:30

So the, there are two types when you come to AI, right? So there are large language models and then there are small language models. So we call it as LLMs and slms. So LLM was particularly meant for, I won't say it's too costly, but it's meant for, it has got lot many parameters and it has meant for like bigger enterprises. And it's not just for bigger enterprises, but a lot of bigger enterprises use this. But when you go to slms, right, it has got the capability what LLM can do, but with some limitations. So I won't say it is as limitations, but it has got its own capability level and which is pretty sufficient for this nonprofit organization to explore what is AI and can reduce their investment AI spend. So these slms, when people say AI, so they just dive into the large language models, which is going to be too costly when they finish their implementation, right? So when you step back and then look what a small language model can do. So it can do the similar things, what we recently did for an implementation for a customer.

Naveen Krishnan

00:22:30 - 00:22:58

So we were initially betting and listening around things around large language models. And then we, when we take a, when you take a step back, and then when we saw like a small language model can do this with like one third price reduction, like for example, if a model cost of 1 million, parameters cost $15, and this can do within three or $4 if you can look at those and then get benefits of those small language models. So it's, it's a win. Solar invincitation.

Donna Mitchell

00:23:01 - 00:23:36

So for those that are interested in coming into the AI space, they're not sure what tools they need, which type of AI they should be looking at, what steps do you think a CEO or a consultant or your high end coaches, how should they be directing or including and utilizing some of the information that you're sharing? But what should really everyone be doing at this point so they're not getting left behind because it's moving so fast. People are kind of confused and they don't know what to do or where to start. Can you help that?

Naveen Krishnan

00:23:36 - 00:24:35

Yeah, sure, definitely. So this is the first point what I say to all my, not just for my customers to whom I meet. So if you are not into AI now and you're not standing out in the market, so you have to, it's a time for you to invest in AI. So think about, think, which is the right one that fits for you because there are a lot of things with AI, it's just not generative AI. So sometimes I have customer asked, like, I wanted to infuse AI in this product. So can you tell me what type of, so you can reach out to your consultants and AI experts and then have them work through your product? Because every product is unique and not all product needs generative Aih. So there is cognitive services and there are a lot of other a solutions which are readily made and available for your use cases. You can even look at that first and then make sure, like if that satisfies your need and then you can go for it.

Naveen Krishnan

00:24:36 - 00:25:13

And as I just said, the SLMs is also playing some roles. So you can look at that. And there are, there are few models that is available online in hugging phase. So hugging phase is one of the open source solutions where you can find models for your specific use cases. You can look at that use cases and those models are available on Azure as well. You can try it from your model catalog and then you can get benefit from those products, those models. So get a clear picture of. So not all needs GPT, right? Not all your implementation needs chat GPT because chat GPT is just one word.

Naveen Krishnan

00:25:13 - 00:25:42

So there are so many other AI capabilities and ML learning. So machine learning, machine learning things are around AI. So which can also go towards the AI implementation. So implementing a chatbot on your application, it's not the end of AI. So there are so many other things which you have to look at. So every product has a different use cases. We can build some different use cases, you can work on these use cases. And if those use cases really need a GPT, then you can go for GPT.

Naveen Krishnan

00:25:42 - 00:25:47

Otherwise you can go for, go for any other solutions that really have.

Donna Mitchell

00:25:49 - 00:26:11

So how does somebody connect or really find out what some of the use cases are or how do they connect with you, or how do they look at what's out there so they can get their thinking caps on and say, okay, maybe I need to be looking at this or evaluating that, or who do they call? What is the next step? If they want more information or want to reach out to Microsoft or to you?

Naveen Krishnan

00:26:11 - 00:27:01

Yeah, sure, definitely. So I think there are a lot of open, publicly facing use cases available which is already implemented to customers. We can go through those use cases and then see if one of those use cases really fits in your product. You can, for example. Right? So as I said, so there is a chart chatbot which is implemented for one of the retail company, and you can look at the chatbot and then you can see how it is designed and how the WhatsApp. You can read through the complete use case document. And once you are comfortable, you can just inherit in your application. And then you, when you start opening just one chatbot, when you start adding one chatbot to your application, then you implement it and then automatically you will realize what other use cases can be built around that and your complete product itself.

Naveen Krishnan

00:27:03 - 00:27:50

If you are staying away and then if you are looking a, then it will be a big mountain when you jump in, really. And then when you start doing some implementations, that's when you will get, get to find lot, many use cases. I have a customer who just came to me with, saying without, I don't know anything about AI. Can you just help me or can you just teach me? So I've written few blogs and it's, it's on medium, it's a, with Navin Krishnan. So go ahead and read through the blogs. And then there are, I have explained like what is copilot and what are different types of copilot, what you have got. And apart from that I can, I've also explained a lot of things about, I write a lot of things about AI, so you can go read those blocks in future after this question, whatever you have, what you just asked. I'm trying to write one blog stating that this is how you can approach AI.

Naveen Krishnan

00:27:50 - 00:28:05

So if you have a use case and if you want assistance to identify a use case, we are here to help you. And this is how you have to grow. And you can wait for that. Within a week, I will try to finish my blog and then release it for you.

Donna Mitchell

00:28:05 - 00:28:32

So that means that you're going to be sending me some links to make sure that we have it on your episode page. We're going to keep in touch so you can keep us up to date on what's going on with you and Microsoft and everything that you're involved with specifically. And is there anything that's happening that you haven't shared with us? You know, I try to get everything I can, all the information so everybody knows what's happening now. So what else do you want to share with us before we close?

Naveen Krishnan

00:28:33 - 00:29:27

Yeah. So what I would say is first try to have some basic understanding about AI. So what AI does and what are the different pieces? And as I said, you can read them, a lot of things available in YouTube, a lot of videos available. There are people talk about AI, go watch them. And also try to read some blogs. Like you can refer my blogs as well. So you can read through those blogs and then get some basic understanding of AI and then see what it can do to you, to your product and you and how it can impact your business and how it can help you grow your business. And once you, once you start identifying a single use cases, just deep dive into that use cases and then try to implement and once you implement one and automatically you will get ideas of implementing few other use cases.

Naveen Krishnan

00:29:27 - 00:29:49

That's how I have ended up helping my customers. So I just help them with just one use case and then they come with to me with several ideas now. So that's how we go build together. So keep in touch with me and I'm available in LinkedIn and you can also ping me. And as I get time I try to respond back all your questions and queries. So yeah, that's how it is.

Donna Mitchell

00:29:49 - 00:30:11Exclude

0:00 - 30:12

This is so exciting. Thank you so much. And thank you for being here on pivoting to web three podcast and everyone who is shaping tomorrow together. And thank you for listening. Thanks for checking in the pivoting to web three podcast. Go to pivotingtweb three podcast.com to download and listen or web three game plan to check out the videos. Thank you. We're shaping tomorrow together.