[Music] [Music] [Music] la [Music] [Music] [Music] [Music] [Music] [Music] please welcome Punit chandok president Microsoft India in South Asia good morning bangaluru how's the Jo come on how's the energy it is uh amazing to see all of you here and uh there thousand plus developers around I don't know if I should call this a room or a hall or a stadium there many more outside and there many more watching live so thank you thank you for being here today is all about you uh you are the developers you are the builders you're the ones who building this country you're the ones who moving this country forward and and we so thrilled to have all of you I'm a coder at heart I started coding late '90s I remember the first language I picked up was C and then C++ I don't know if anybody codes in I remember the book was uh the book was called letter C uh people remember it okay I was thinking I'm in a different generation uh but I remember the first program I wrote it was a stupid game I made but that day my world view changed I felt I can build something right uh and I and all of you build every day so we're so glad to have all of you here um and by the way looking at the days of let us C and let C++ look at where we've come with AI um if you look at the picture behind me it's a absolutely outstanding view of Bangalore or bangaluru as we know it and in the future that's coming our way this is made by co-pilot I asked her to build it last night U I said I should wait welcome you in bangaluru style uh but we've got a really exciting lineup for you today folks um I'll bring bring in Satya on the stage in a few minutes and I know you're all waiting for him yes uh Satya is going to be here in a few minutes and he's going to talk about how he AI is going to shape not just our lives and our organizations but our societies and our country and I'm really looking forward to that and after that I have a few of my favorite people coming on stage Rajiv Kumar who runs IDC the Microsoft IDC in India charumati shasan who runs all of our cloud and AI capabilities in India and sandep alur uh who runs the Microsoft Technology Center in India we going to come and have a real conversation around how AI is going to change what you do every day how is a AI is going to change engineering development every day with that um by the way I also want to welcome a set of people who are not in the room can we roll the video please and I said I'm amazed to see 100,000 people around in the room today but we've got 2 million professional student developers across India who are engaged with Microsoft and everything that we're doing in Ai and today I'm so proud to tell you that 100,000 developers from across the country from 100 Microsoft developer communities from 100 cities are watching this live today so the three hunds 100,000 100 cities 100 developer communities welcome to all of them welcome to all the people who are live streaming and joining us we're so glad to have you you're the ones who make India the AI Powerhouse that it is so with that uh let's get going um by the way I have a confession to make I'm relatively new at Microsoft I'm 5 months in so I'm still trying to find my groove learn every day so I needed help to pull this event together and I went to my colleagues for help but I also went to co-pilot for help and I asked co-pilot what should be most interesting for developers who are coming in to listen to Microsoft and what's happening in air what are the topics that'll be top of the mind for them it gave me a list of four or five really interesting ideas but the three that I picked up that I will talk about today number one what makes India Unstoppable what's really going on in India today what's the shift that we seeing every day around us number two what makes AI Unstoppable what's really shifting and Satya will go really deep in this as well and then finally the most important part what can make each one of you each one of you in this room and the 100,000 people who are logging in Unstoppable together right so that's what I'll attempt to do let's get going with India um my favorite quote on India is by a guy called rer Sharma who said India is a country that consistently disappoints both the pessimist and The Optimist India is a country that consistently disappoints both the pessimist and The Optimist but that is changing now we're only disappointing the pessimists and I'll show you why and there are lots of statistics on this page I'm not going to go through all of them you know these numbers everybody in India knows these numbers but I'll just pick three for three of them quickly for you number one we know we we know India's Got 100,000 startups we're the third largest startup ecosystem in the world but do you know India is producing 100 new startups every day 100 new startups are coming up around you a lot of them in bangaluru and Hyderabad and different cities but 100 new startups which is more than any other country in the world and that's an outstanding number every sixth by the way that's worth a clap for sure thank you I'll give you another statistic which is worth a clap and I'm kind of edging for claps now but every sixth AI developer in the world today every sixth AI developer AI researcher is coming from our nation every sixth developer it is just outstanding to see what you all are building um and finally the last statistic that I'll give you which is U on the bottom of this chart 25% of the incremental Global Workforce 25% of the incremental Global Workforce over the next decade will come from India now if that's not impact on the world I don't know what it is right so it's incredible but more than these numbers there's something else that's shifting in India today Indians are dreaming big all of you are dreaming big like never before and more than just dreaming big we're all chasing our dreams like our lives depend on it like our lives depend on it and I think that is the big shift that's happening and this will lead this generation which is all of you this generation of Indians to redefine everything redefine everything from work to success to morality to technology and I think that's the shift that's happening in India which the world is now starting to notice India matters to all of us and I can say it for a fact I can look at your eyes and your faces India matters to all of us but we now want matter for India we now want a matter for India and I think that's the that's the unlock that this country needs and it's coming okay so enough about India let's quickly talk about Ai and uh what you see on this chart um is a set of customers in India you know all of these logos we work with them every day they're all building on on Microsoft they're building AI on Microsoft and our purpose in India as Microsoft India is to be the most trusted the most Innovative AI partner for India AI for everyone that's why we get out of beds every day that's what we trying to do every day and we've moved from talking about AI to applying AI at scale and customers always vote with their feet we all know this and if you look at this chart there's a March going on towards Microsoft when it comes to Ai and by the way these are not just uh these are not just uh logos there are real stories behind them and uh we've got lots of material outside at some point when you're walking out you should look at this but there are lots of showcases and use cases that we pulling together of real impact let me come to the final part of the conversation India is becoming Unstoppable AI is getting better and delivering real impact every day how can all of you become Unstoppable and I'll give you three simple tips I'll go to First principles I'll keep it very simple the first and these are my three calls to action learn Ai and this is my cute take on AI IQ I call it aiq pardon me for that um but I genuinely believe that the opposite of knowledge on AI today is not ignorance it's not ignorance it's the illusion of knowledge sometimes I feel I'm trying to figure this out and I'm there I'm not there by a mile we all have to learn and if you're not learning every day and by the way that's something you should teach us you guys learn every day but if you're not learning about AI every day you were you are falling behind don't do that learn every day number two unlock productivity start using co-pilots and I showed you the picture I showed you co-pilot helped me pull this agenda together and my own personal productivity my email productivity has gone up by 50% two3 of my emails I don't write anymore a lot of my meetings get transcribed and the action and the summaries come together beautifully it drives a ton of Effectiveness in the US we seeing 30 minutes of time saved per user per day imagine if all of you get 30 minutes of time back back every day to things that really matter to you don't miss the opportunity and by the way AI as I said right start I started by saying learning you have to learn but learning is only abstract unless you really put this into practice and finally Drive AI capabilities in the organizations you work with start don't don't just make sure that people around you the ecosystem the organization are not just dipping their toes but they're actually using AI so I'll keep it simple learn it use it deploy it learn it use it deploy it and that's how we get going and then of course we need to work on data and governance to bring all of this together let me wrap this up and I'm going to leave you with three messages quickly folks India is not just incredible it's also credible now it's the most exciting Market in the world today and it's a privilege to be in India today as Indians building for India AI is truly becoming Unstoppable it's real now the impact is real and finally all of you can be unstoppable if you learn it every day if you use it with co-pilots unlock productivity and get started in your organizations and move everybody around you to get going with with momentum with that I'm going to quote the man I'm going to bring on the stage um he said our industry does not respect tradition it respects Innovation it only respects Innovation and there couldn't be a better time in India today to innovate with AI and with that it's truly a privilege for me to invite on stage Satya nadela chairman and CEO of Microsoft sorry S I didn't see you welcome thank you thank you thank you so much uh Punit it's wonderful it's wonderful always to be back in India and it's obviously wonderful to be in a room full of developers um you know I always remind myself um that Microsoft at the end of the day is a developer company we got started building tools uh and here we are close to 50 years since our founding uh as excited about building tools and platforms for developers you know having lived through my 32 years at Microsoft four major platform shifts uh PC client server uh Web Internet uh mobile cloud and now the fourth one being AI um one of the things as developers I think is super important for us is to be very grounded in what really changed what's new what is this platform shift about and dimensionalizing it um I think even last year when I was here we were all excited about chat GPT had just come out and you know we had lots of fun about sort of what all it could do and in a year's time things have progressed and I think even a we have a better understanding of what are the two dimensions of this platform shift that I think all of us as developers have to one fundamentally stood of really you know deeply understand and then take advantage of if you take the 70 years of or so of computing history one of the dreams has always been can we create computers that understand US versus US understanding computers right if you can go back right one of the things even even today when you go back and read maniva Bush's paper uh as we may think uh or read any paper by angle Bart that was it that they were really really trying to get computers to understand US versus us trying to understand computers and I think we finally may be there uh with this breakthrough of natural language or models that understand natural language computers that can hold a conversation because after all a conversational interface is ultimately the easiest most natural user interface the fact that it can be multimodal text images video in and out uh the fact that it can be arbitary length conversations the fact that you can have multiple domain conversations right I can talk about about cricket and I can talk about uh baseball and it won't get confused uh all of that I think does allow us to actually have a very different approach to user experience uh so that's I think the first one then the second thing is you know one of the fun fundamental things we as developers do is digitize right we digitize people places and things right we put it into Data models and then we query it we reason about it uh That's essential like all of Computing is fundamentally getting up in the morning and digitizing and then trying to make you know really build insights and analytical power and predictive power from that data uh now we have a new tool in our Arsenal uh we have um this new reasoning engine think of it as a neural reasoning engine right which can find patterns give us that predictive power right when you think about a continuation a summarization all of these are new things that allow us uh to make sense of all of the data being digitized so I I feel that as Developers for us to really understand the power of these two platform shifts or these two capabilities that are part of this platform shift I think is perhaps the most important thing and that's why uh we're you know are you know and all this is obviously at the end of the day only going to be worth it if we can take this and in you know as input that drives economic growth in India right to me uh that's what grounds us even as a company um uh you know to be able to say how can this AI contribute to economic growth in the United States there's some fantastic statistics that that something like uh 10% of what is going to be uh India's GDP growth can come from the output quite frankly of all of you in the room uh and so that I think is what's at stake and that's what uh grounds us in our mission to empower every person and every organization in India to be able to achieve more as past and really take advantage of this platform shift that means empowering developers in India to be able to build the products and services and scenarios not only for India but for the world and that's what's exciting uh and we for that reason what we've done is basically taken this platform shift and etched it into the length and breadth of every layer of micros Microsoft uh stack people ask me hey what's the AI product I'd say everything uh what we're doing whether it's in the infrastructure to the data layer to the tools every part of what we have done as a company now thinks about AI as first class it's not a bolt on from first principles how do we build like in fact you know I having sort of worked in our data center business for now close to 15 years I myself cannot recognize even the core system architecture of what's powering uh you know these clouds today is so different than even the thing that you know I started with maybe 15 years ago and so it's fantastic to see how from a first principles basis we are able to change uh every layer of the stack now the opportunities ultimately when I I had a chance to even see a lot of showcase applications that are being built uh you can change employee experience you can change the customer engagement you can change every business process application across every industry and most importantly the curve of innovation in any organization uh can fundamentally be bent and I think that that's the I would call it the transformational opportunity uh that is really in front of us and that's what leads to the three imperatives for all of us that I wanted to talk about the first one is all about being able to use these co-pilots right ultimately the idea is to be able to use a co-pilot to really help us Empower ourselves to be able to do more now I love the design decision we made back back in the day to say let's put human agency and humans in the loop with powerful AI so that we can achieve more right that's is is as much of a design decision um and that pattern I think has really picked up in fact the first time I fell in love with this was when I saw I think GPD 3 uh it was a it was called something else we had a internal code named for it it was actually a derivative of GPD 3 in the context of GitHub uh uh and when I first saw GI up co-pilot is when I sort of felt like wow the age of these llms has really come a long way and the fact that we didn't even train it uh quite frankly on code uh and it sort of learned uh coding was the fact that it had scaling effects working and emerging capability and that's when I felt if you put that into a right design metaphor uh we can achieve some real big breakthroughs and that was the Genesis of get up co-pilot and so and of course now we have the micros of co-pilot and co-pilots by business process for sales and service and for security and what have you and really these age of co-pilots has taken off uh in fact we're going to launch a new ad uh as part of Super Bowl this Sunday uh in the United States so we thought we'll just roll it out for you to just have a take a look at what the age of co-pilot could mean go ahead and roll it good [Music] morning [Music] [Applause] [Music] watch me just watch me now just watch me now I got something for you a little more for you you won't believe your eyes just watch me now I got something for you little your eyes just watch me now just [Music] watch you know it's it's very exciting I I remember when I first saw chat GPT um it was sort of like you know seeing Mosaic for the first time in 93 um and similarly I think when I look at this co-pilot you can see all of these software categories that we know of operating systems browsers productivity software what have you IDs all of them can collapse into this one user interface which is essentially a co-pilot that can be invoked by your intent uh and really not just give you the answers but to create to be able to take action so we're at the beginning of a very very exciting Cambrian explosion uh when it comes to software Innovation and also a grand collapse of a lot of the software categories today and so a tremendous opportunity and you see the productivity stats by the way already I think Punit was talking about his own personal use of productivity uh and how much time it saves it's knowledge turns Andy Grove used to talk a lot about this thing called knowledge turns inside organizations everything gets faster uh what's the power of AI it's kind of like lean lean brought to knowledge work uh I feel that anything that you do you can do it though much faster can it be coding can it be responding to an email making sense in a meeting things that I think we spend a lot of time on if you only you got faster uh what does it mean to our organizational output so it's an exciting time uh for all of this and lot lots of uh customers right here uh for all these co-pilots in India it's tremendous to see uh of course this being a developer conference the thing that we want to talk most about uh is get up co-pilot is been revolutionary you know nowadays I can go into my weekends and actually start a Dev project and complete it and especially the combination of code spaces and uh get up co-pilot as being a GameChanger uh setting up environments used to be such a pain and now it's just such a joy because somebody else does it for you uh but overall I think uh that the productivity of developers is something that we care deeply about um and in fact uh today the uh the customers that we have some of the best big biggest startups whether it's swiggy or misho are using it in fact I had a chance to meet with the open Healthcare Network folks that's a community that was born on GitHub it's fantastic to see what they're doing right so they're built an open- Source EMR system they even built an open source Telly EMR system uh it's just superb and it sort of obviously composes with the rest of what's happening in India which is so unique around these digital public goods uh and the fact that they're now available in 200 plus hospitals across whatever nine states in India uh and it all was a pandemic project that's just taken on a life of its own but the cool thing was they all use get up copilot everybody who is part of uh the uh the open source community building this is using get up co-pilot and it's helping them by the way bu do the code uh definitely do the testing uh and then most importantly super helpful with docks uh and that's what's really making the community sort of that much more productive and it's fantastic to see that HCL for example is using get up copal but they're also integrating the work flows uh from between the developer workflows and the Microsoft 365 co-pilot so there's lots of lots of innovation uh I think that's happening around co-pilots and it's exciting to see India is a place where you know what the momentum around developers and development is unbelievable like in today as we speak India is number two only next to uh the United States in terms of the total number of developers on GitHub and it's going to cross in 2028 I believe or 2027 uh it's going to cross uh so it's going to be the place where the most number of developers are going to be in uh second even the AI projects I I mentioned this even the last time I was here I was pretty stunned when I saw the stat number of gen projects on GitHub uh with Indian contributors is again next to the United States so it's just awesome to see the human capital of this country uh really Embrace this new platform and effectively lead the new platform shift and it's wonderful to see that uh so congratulations to all of you [Applause] all and and and to sort of really talk about that right which is what does it mean to have these tools not just be things that you use that were built in the west uh but tools that work uh for all of you and really give you know I would say that Indian flavor to it uh um so one of the things that I'm very excited about is some of the indic language and GitHub co-pilot coming together and to show you all of that and more I wanted to invite Karen up on stage from our GitHub team to do a quick demo of giup [Applause] [Music] Karen thanks hello everyone really nice to see such a large audience uh eager to know about GitHub copilot and what it can do for you all right so today I'm going to show you three things getp copilot can do for you explain code that's already been written create code you need to write and also collaborate with others when creating P requests so let's get started all right so here I have a website I'm building with typescript using giup copilot and I want to add some signup functionality to this so let's first add a signup button I'll write a comment for it and get up copilot writes the code for me working with me along the way so all right there it is now I want to make sure my customers are protected with secured passwords and I need some guidance on that so my colleague gave me a simple password validation code to start with but it's a regular expression and hard to read with giup copilot it truly feels like I'm working with a pair programmer and in this case I can use the slash explain command with kup copilot chat to help explain code that I'm unfamiliar with even if it's a regular expression and here's the amazing part I can even do this in in Hindi by asking it an equivalent of what does this regex do so I can ask it reg here fantastic right so you see how get up copilot picks up its cues on how I would like the answer based on how I asked the question now get up copilot chat can also be invoked in line so it can work with the code in place so to get to it just hit control R command I now I'll ask co-pilot to help me write the entire signup page and this time I'm going to again Ask it in Hindi but this time I'm going to notch it up further and ask it in the DEA nagari script all right so I'm going to ask it a English sign up page B email password or password confirmation M and amazing so you can see that get up copilot didn't just understand what I wanted in Hindi it's actually written the entire code for me so I'm going to accept that change and like a good programmer I'm going to add one other small thing to make this work but again using GitHub copilot so I'm going to go back and then see hm okay so I can keep iterating on this website by highlighting code and asking co-pilot for a change so uh this time I'll ask it to style the form and I'll also ask it to solve the second hardest problem in software development centering a div so uh but this time I'm going to write it in Telo script all right so I'm going to ask it t CSS style and so on it's going to be gentle and then ask it and let's see nice so it gives me it Styles me and uh that looks good so I'm going to accept that change go back and yes all right so let's get a pull request created and submit it to my team for review but like me GitHub copilot is not just in the editor it's also with me on github.com so giup copilot can help generate a description of the changes in the pull request and it doesn't just list the changes it also gives you useful meaningful description of what the change actually does and the best part is that it also gives you deep links to the lines of code that has changed that's it I've shown you how how your patient multilingual pair programmer can help you explain code create code and collaborate with others so you can build more faster thank you I tell you regular expressions in any language are hard I mean now weal slash explain I tell you nightmares um you know it's so fantastic to see uh the innovation and get up co-pilot I think the thing that he showed in the last uh you know was just awesome right to me what I think where we're going in fact at get up Universe we showed the last demo uh I think that's the future where start we don't start just with the co-pilot but we start with a PR uh in fact we start with an issue uh and then it generates the right PRS for everyone else to be able to then use copilot to complete the task and I think that's sort of the agentic future uh that we can look forward to it it's going to be a tremendous World um look I think the next imperative of course is let's say we have the best tools right here we are a software developers we have the best tools then it's time to change the world that is build your own AI applications so that's the real uh you know I would say the exciting part of this platform shift now to do that one of the things we are doing is taking effectively the stack we have used underneath GitHub co-pilot Microsoft 365 co-pilot underneath chat GPT and making it available to all of you right every layer of it starting right at the infrastructure uh we're building out the data center footprint as I said what is a data center region or an Azure region today especially with what's happening with sort of the AI accelerators is unlike anything I've seen uh everything from the power draw to the hwx system to the liquid cooling uh is going to really change the landscape and the other exciting exciting thing for us is to sort of really really have that sort of spread around the world in in the in India we now have four regions or we have three regions forthcoming uh and we'll keep expanding uh so that you have the compute resources you need uh we're also uh building the best AI infrastructure so these AI accelerators for training large models uh one of the things is these are very interesting workloads right I you know having sort of lived through the Big Data age uh and I thought oh I understood data parallelism but these are synchronous workloads uh that really are low latency uh and so therefore thinking about the memory hierarchy uh very differently uh thinking about obviously the the Silicon very differently and that's what we're building for large training and inference and small batch training so we have uh by the way there's fantastic Innovation happening all the way from Silicon up right so this is a great time to be in that combination of a hardware software engineer uh where you're thinking about system architecture the other thing that's exciting by the way is mors law is fully working uh mors law is absolutely working when it comes to AI accelerator so that's another area where I think anything that's sort of super expensive today you can be assured uh that the cogs are going to come down so that one of the things that I tell our developers in Microsoft is be super ambitious today about what you can build with the best models knowing that in 18 months time it's going to be oneth 100th the cost of what it is today and that's you know very empowering thing there's no other input in the world that has that type of price characteristics uh so infrastructure is one AI infrastructure then of course the models themselves right these are new platforms uh we have the best large models with the all the GPT varieties GPT gptv gp4 turbo all of that's available to you uh we have all the open source models in fact whether it's mraw whether it's llama uh we want to make sure that the best open- source models are available uh we also have a model catalog with everything from the hugging face on uh so exciting time to be able to access all these models the other thing that I'm very very pleased to see is that you know this is not the last of model architecture breakthroughs there is real new breakthroughs even happening in model architectures um and to me to see what's happening with Fi or small language models it's one of the most leading models for when it comes to small language models that Microsoft uh built and you know it's a different approach to even training right so instead of thinking of training on all tokens what if we could train uh on these reasoning tokens right the original paper was attention is all you need which was the Transformer paper and the paper we wrote was textbooks is all you need uh and it's sort of intuitive right so when you want to learn something new uh you don't go and say let me read everything about it I'll read the textbook and so the idea is if that applies cognitively to us it should apply even to AI training and so we'll see how it is you know how it'll prog rogress but I'm very very bullish uh based on the results we have seen today and so we have a very very aggressive road map to keep advancing uh fire you can run F obviously in the cloud you can run it on your phone or a PC and so you'll see a lot of great stuff happening in the slm space as well uh and I'm excited to see sort of new Innovative startups right here in India in fact I had a chance to meet uh uh with the founders pru and vi from saram Ai and um of course PR worked at MSR in India uh before and we're excited to support them building out their um you know llm which is trained on all the indic languages in fact I the demo they show me is tremendous it's kind of like taking that voice mode that you have today in chat GPT and making it available effectively for any application developer uh to be able to have as the front end for their uh application in India with all of the indic languages is fantastic to see and one of the and they have unbelievable scenarios that they're building out so let's just roll the video to give you a flavor for what they're up to running any business is hard but it's impossible if you don't know what your customers want what if you could talk to your customers easily hello go deeper take feedb in their own [Music] language with you can understand thousands of such conversations in mere minutes read between the lines and understand what people are truly thinking insights that can change what you build and how you build it it's awesome it's exciting to see I mean you can imagine right so you can take something like GPD 4 you can daisy chain it with saram and then create these types of applications uh which could have fantastic impact uh in any Indian language and the mix of Indian languages um you know the other area of model development that I'm very excited about is not just about human text or human modalities uh but also trying to learn the language of nature uh because that's another Ultimate Dream right so if you're going to really tackle say one of the hard challenges in front of us which is the energy transition you have to take something like 250 years of development of say chemistry um and compress it into 25 years uh and that means you got to really take what we've done with let's say hpcs and so on with simulation uh and accelerate it and an AI can absolutely play a massive massive role in fact I think about this right if you sort of thought thought about hpcs as simulation layer um and in fact Quantum ultimately will be the accelerant to that simulation layer think of the AI layer as the emulation layer of that simulation right so think of it as reducing the search space helping us let's say find a denuo molecule uh for material science uh and one one of the things that we're very excited about is say something like maten which is uh a new model that we have which we used in order to develop some new materials in fact we round tripped this we worked with one of the National Labs in the United States came up with a set of molecules uh fabricated a new battery which reduced the lithium content by 70% all right so that so it's absolutely possible for the next startup in India that goes out after uh batteries uh for let's say EVS or what have you to come up with new material signs essentially to change uh what the assumptions are uh and I think that if you think about the hierarchy starting from cells to Mo to to materials all the way to biology biology being the hardest we can start making real breakthroughs with some of the same models that we have used for language uh of course we will also have different model architectures one of the other models architectures is you know why should we not give some inductive bias to these models so for example a graph forer has a graph that has knowledge and a transform or bolted on and so those you know that's the thing that we're using for a lot of the biology stuff so I think there's going to be a tremendous amount of model Innovation uh and all of these will be available to you like any you know uh large language model as well and so science and in fact I know in India now we have sort of some of the core tech companies that are coming up in all of these fields and so it'll be exciting to see what all you can do uh in accelerating uh science of course when you think about the stack the data layer right with models is fantastic you have the infrastructure you have the models uh but the next layer is data uh everything about the data layer has to be rethought where there is not only the traditional stuff right you run SQL queries you can run spark uh but you're also going to run these large language models on top of that same data so that means thinking about whether it's a vector indices or what have you built into your databases that's what we've done whether it's in postest whether it's in SQL or in Cosmos DB uh we have Azure search which is what is in fact the retrieval inside of chat GPD is powered by Azure search uh so all of these are again Azure services that you can use inside of your applications whenever you need an embedded Vector space uh so that's sort of one side of it the other thing that I'm very excited about is fabric right one of the ideas we always have was how do we change or disaggregate what is considered a database today so there is the one Lake storage system there is a computational system which by the way you can have SQL spark or Azure open AI all three are just compute jobs uh and then you have a business model as a developer building an application I can sort of you know mix match whatever compute on top of all of this on one reserved instance if you will uh and so I think this is going to have a profound impact on how we think about even the data layer in conjunction uh with the rest of the application so that's another one and of course the the next up is the tool chain uh as I said we're a developer company we're a tools company and we're building out the tools whether it's GitHub vs code on one side whether it's the low code no code tools with Power Platform bringing it all together uh with Azure open AI or Azure AI as a tool chain uh which has access to all these models uh and AI Ops being built in and so on so definitely the tool chain focus is a big big deal for us and you have now the access to it and you know it is fantastic I had a chance to meet uh a bunch of developers who went through 100,000 of them who went through this U Odyssey initiative so [Applause] congratulations it's great to see all of you picking up new skills uh because that's the currency today of being able to take uh your AI skills and then converted into projects that you're doing uh whether it's in school whether it's at work whether it's at a nonprofit I think that's the impact uh that I'm really looking forward to seeing in the years to come um and again in terms of the applications being built they're built you know I I had a chance even yesterday in Mumbai to see uh some of these uh demos of uh you know what Air India built in terms of an a agent um you know it is very very cool in fact it's one of the best agents I've seen in terms of being able to not just get all of the travel planning done but to take action uh right there and so it sort of tools use uh inside of a bot is what they're doing uh it's great Arvin which is one of the oldest companies uh in uh India is you know think about the diff rate of diffusion so software developers working in that company uh were able to sort of build a contract management system for their legal department but more interestingly one of the things in this age is that it's no longer about just information at your fingertips but it's expertise at your fingertips so it's not just the legal department that now has a bot or it has a co-pilot for being able to make sensor contracts but anybody in the company especially in the commercial side of the organization can also use these co-pilots to make sense of contracts right so that's the type of workflow change that I think one can expect uh jugle Bundi this is the thing that really flawed me the last time I was in India uh it was fantastic to see the progress they're making again Microsoft research along with the jugal Bundi team built a low codee no code tool so that any nonprofit organization can start using these llms to build new tools um and I think there's a there's a video uh or you know this is the ambient of what uh you know you can do in terms of building out these tools is fantastic to see uh the the nonprofit organizations embracing it and I also had this morning uh the chance to see Sika uh which is being you know built out uh by again a nonprofit organization right here and the beauty of their approach is everybody talks about students we know that I think one of the dreams of being able to deliver personalized tutoring uh to every student in the world is absolutely something that we can do but one of the things that I always think about is really empowering teachers because at the end of the day if we can put tools in the hands of teachers they have the ability quite frankly to inspire uh more students than any AI can uh and so to me uh really having uh Sika uh really be that tool that allows teachers to be able to do their job more effectively so let's roll the video sea Foundation started as an NGO that's focused on the education sector we are working in Six States across India in about 50,000 schools impacting nearly 3 million students the public school system in India has been severely stressed we found that typically a teacher handles multi-grade multi subjects especially in the rural areas of the country the biggest challenge was creating lesson plans preparing those lesson plans it was taking a lot of time everything was manual we were writing and writing and writing and that's where I felt technology can help shika co-pilot is a generative AI solution where we want to empower teachers to create engaging content for their classroom these teachers are able to generate lesson plans they're taking the activities that are generated and they're using all of these cont content in the classroom within 5 minutes or 2 to 3 minutes I'm preparing that lesson plans it's very essential that any solution we roll out supports the local language is based on azur open AI it's grounded in local context the problems that are faced are pretty much Universal so if we have a solution that works in just one state in India you think that solution can go [Music] globally you know to me when I see that video and when I had a chance to this morning to meet them the thing that sort of really inspires me is that combination of folks coming together right which is Microsoft research India uh a nonprofit organization and the teachers all co-designing it together I mean that's when you have that type of feedback loop creating new products that's magic and that I think is at the end of the day what what we all live for and it's fantastic to see that so that brings me to the last imperative which is of course thinking about about security uh and AI safety as First Class part of our project as opposed to something that we do after uh it starts uh really with security uh one of the things when we think about building powerful applications powerful new models what have you you got to have a very robust security architecture which is zero trust uh everything from the endpoints to Identity to your application and infrastructure and this is a a huge huge Focus for us uh when it comes uh to really building out our Security Suite uh so we're excited about that so that's I would say the first layer and it's not just the having the products uh and the security layer but to be able to have the operational security posture uh that allows you right as I always say you can't get fit by watching others go to the gym you got to get up in the morning and go to the gym yourself uh and that's what really uh I think security is all about uh the other aspect of it is um the safeguards um to all of your data because the core asset in many cases is going to be your data uh you want to be be assured when it comes to whether it's a software developer or an Enterprise our promises are very clear we're not going to use your AI to rather your data to train any large models uh your data you know and your data is your data your AI models are protected again using that security architecture uh and then also copyright uh one of the things that we know is that that all different countries will all come up with the copyright regime uh but we do need an enlightened copyright regime and in any case we want to take the risk on on behalf of developers and mify you so that you can use all of this because at the end of the day uh there is fair use that's what has help Humanity get ahead uh with transformative technology so you can't have a overburdened copyright regime but at the same time copyright needs to be also be protected because anybody who is producing content uh that's copyrightable should be copyrighted and so the thinking of even training and fair use uh and then uh what is copyright protection is all something that I think we'll all get settled in the years to come uh but we definitely want to make sure that this it doesn't stall The Innovation um the last thing I would also say is that when we talk about AI safety we've you know this is something I wrote up piece and slate I think in 2016 uh and since then we've taken those principles translated them into a set of corporate standards and set of engineering practices quite frankly so it's not just a set of slides or a memo but it's really at the end of the day you shift left have a set of engineering practices that I'll think about AI safety in fact when I talk about Azure uh AI tooling the most popular Services most important services are all the safety Services right that uh we built we built for Bing we built for chat GPT we built for co-pilot are now available to all of you as developers to be able to use as part of anything that you do in order to have the classifiers in front uh of your applications and what have you and then it's not just that but also being able to do audit right so just like anything else like quality uh has a test function what's the audit function on AI safety so that's again an approach that you all inside of your engineering practices can take around AI safety um so these are the three real imperatives and hopefully that gives you a flavor for what are uh what's ahead of us in terms of the opportunity as well as I think the tools the platforms uh that we can exploit to build any application that you can dream of um you know so that's what grounds us ultimately again I want to end by uh you know where I started which is our mission to empower every person and every organization uh in India starting with you all as developers uh to do more and um you know one of the things that I'm very excited also is to bring one of the initiatives which we've had called code without barriers uh and expanding to India and really helping 75,000 women developers by [Applause] 2024 but one of the things is it's not just about skills right it's about being able to create Economic Opportunity for an increasingly skilled Workforce uh and in that context I had a chance uh yesterday to meet uh the founder of karer yesterday along with one of the beneficiaries of what they're doing it's just tremendous and again something that incubated in Microsoft research it is awesome for me it gives me great pride to see work that happened right here in uh Bangalore uh in Microsoft research now really become the foundation uh of our ker and what they're doing is taking AI jobs uh effectively and part of the AI economy and taking those jobs all the way to rural India so I want to leave you with that video have a fantastic rest of the conference thank you very very much India has 22 official languages and maati is actually a language that is spoken by millions and millions of people but unfortunately many Indian languages are under resourced we cannot let language be the thing that does not allow you to use technology AI is a transformative technology people deserve to have access to these amazing Technologies in their language Cara exists to accelerate social mobility in rural India by building AI based solutions to help our communities our workers record sentences in their mother tongue on their smartphones and for the simple task we pay them nearly 20 times the Indian minimum wage and that leads to a data set so they keep on making royalties every single time those data sets are resold I think it is imperative that if we are changing someone's income structure in this way that we also need to help them understand financial literacy AR K's platform is hosted on Azure we use Azure cognitive services for our own AI based validation work and we also [Music] using Mal if you can solve it for India you can solve it for the world please welcome sandep allur CTO Microsoft Technology Center India wonderful good morning everyone that was such a profound statement in the end it's great to be here are you all pumped up yeah wonderful we'll do something very unique this morning by the way I have the unique opportunity to host two celebrated L from Microsoft they each of them spent 30 plus in in the 30 plus at Microsoft by the way they developers at heart and I thought before I welcome them um we'll do a bit of a unique introduction for them yeah are you with me on this okay so before we get started quick question we saw a wonderful demo from Karan he tells me it's G you can use it how many of you going to try a multilanguage conversation with copilot today after this ask in Canada Tamil telu it's fabulous right it's great I don't know how it looks but should be fabulous so before we get get going the conversation um we've used a term called co-pilot all these Ways by the way who is the pilot yeah right imagine I'm the developer I'm the I'm the pilot right so I thought I'll ask my co-pilot to give me a nice song lyrics for the theme which is I'm the developer I'm the pilot yeah but we going to make sure we never say the statement when the real pilot is there yeah because we are truly the yeah let's do that so my question to my co-pilot is give me that song Let's see what comes up should be very interesting it uses sunno plug-in by the way it's there and you know co-pilot Isn't that cool looks right it's a w thing right when I saw that for the first time it's like this is really cool yeah there's a wor there's a chorus and look I'm a developer I'm the pilot in the world of zeros when I'm flying hi it's really really cool um but there should be music correct it's just lyrics I said we need a music I thought we'll bring music and welcome my two guests are we ready and again this is my request I'm a local boy I know how crazy we as developers in this city we make loud noise right and when the music comes they're coming on stage let's give them a bangaluru welcome can you do that okay wonderful so when the music begins let's cheer them up my first guest um she is CVP for engineering and 32 plus years in Microsoft and one of the most celebrated leaders and we all look up to her put your hands together when the music comes okay and welcome Chu shasan let's [Music] go I am the Creator digit command now through the feel it's say oh right doesn't no it's literally It's Magic right we all know the fact that this innovation in front of us can see hear and speak and look this is magical now let's move forward this is a chorus right so now once again need your support to welcome uh my second guest again he spent 30 plus in Microsoft he heads Microsoft India Development Center MD for micr Development Center please put him together and welcome Rajiv Kumar [Music] [Music] Raj perfect you guys like the music loved it how did you create it my copilot yes awesome let's get started now Now quickly um by the way so much is happening we saw some super cool Demos in satya's keynote as well um what inspires you the most when it comes to the current Innovation Char let's go with you first great you already dated me uh started in the company the early '90s I've seen three of those big Paradigm shifts that Satya also talked about um and the latest one the AI paradigm shift is just mind-blowing and to me I think what's exciting is the rate of diffusion right if you just go back and think about the fact that these AI models are now transforming lives of rural Indian farmers and Children Learning in Indian schools uh I think Tua we just talked about Tua and serum. a it's just mindblowing the rate of diffusion of the tech and I've been through multiple of these spadam shifts in my career three of them precisely in the fourth one is ongoing and I really am just amazed at both the pace of innovation and the other part that's super exciting for me uh suep is the fact that we're in India all of us and India is not just is is in many ways there's really no rate of there's no delay or impotence mismatch between how India's adopted Ai and so India is actually helping scale Ai and so to me that really is another pretty fascinating story and uh and all of you are are part of that magic so I'm super excited to be here and be part of this massive paradigm shift that Microsoft is leading for the world awesome rajie your point of view well I I think can you hear me okay okay uh I I'm today I think as as excited about the new platform that you heard about as I was 30 years back when I joined Microsoft when I joined Microsoft I built the visor basic compiler and now I build co-pilots and the journey has been just the excitement is absolutely the same so looking good perfect awesome so Rajiv um we'll move to the First theme which is basically Raju andaru um the power of the current breed of innovation right I thought I'll bring in a very short demo uh and then we'll we'll pick your brains on how is developers will prepare that for that particular stream of innovation yeah so this is a little experiment I did this um over the weekend you can switch to my slides um by the way I'm using um Chad gbd plus here right um how many of you have chat gbd plus access here just checking a few of you that means you can try it to yourself um I said earlier this Innovation can see here and speak I wanted to experiment and here my question to the model is I want to change a bul please help me I'm showing this particular visual here in a few seconds it's going to you know come back and assess the situation and um it'll basically tell me that you know what you basically have to do counterclockwise to open this particular lamp shed that's what I highlight right there and I do that and I say you know what cl cl has worked now tell me what next I need to basically put this bulb and of course I didn't upload what I see if after I removed the particular lamp shade now it says you know please share sh me what's inside I'm going to share the socket visual here and um you know as you see there this Innovation is able to see things and tell tells me that you know what um it's basically a standard screw based helps here right and I I need to know that okay now I have two bulbs which one should I basically put here right interestingly right it's a by the way one conversation here I didn't even do anything else it tells me one on the right yeah that's Isn't that cool that's right yeah this is the power of magic so right so yeah hats off to people who build this right now with this context right so now um Char question to you developers here uh you know when it comes to building applications always been API first approach we're shifting towards AI first approach right so do you think it's time for our developers to reimagine how you build application with AI infusion absolutely and uh I think an AI first application is something where you don't think of AI as an afterthought you build it into the core of your application and the way you do that is you really start thinking about certain things from a very fundamental perspective right and I think Satya did U amazing Justice to this topic in his in in the keynote and the entire kind of tool set that the developers use is going to have to fundamentally shift and uh one if there's one takeaway I'd like you folks to walk back with is that really go back and think about your data and information architecture because that is at the end of the day the currency on which the AI operates so uh Satya talked about the one Lake the fabri Microsoft fabric one Lake really thinking about how do you bring uh unstructured data structured data and you bring realtime data into into like a one L Type architecture so then you kind of build models and how many of the folks in the audience are developers hands if you can raise your hands uh what about data scientists yeah I definitely see that so interestingly this world is going to massively shift the way developers think about their jobs we heard about the massive productivity increases uh in one study we found out that um with AI pair programming you were nearly 55% faster to kind of spit out a uh a j script kind of application right so sanip I I really think at the core of this is that productivity gain right for developers it would be about completing code about unit testing Automation and we seeing massive shifts at Microsoft my own team is is beginning to adopt this at a pretty rapid scale and we're seeing that massive productivity gain but for the data centers out there uh the models that you create the way you're able to build those models and test those models and kind of put them out there I really think there are massive again platform capabilities the Azure AI Studio as an example you can uh there's models as a service so this I think really rich capabilities in the underline platform and when I reflect on like how we used to do work in the API first world versus how how you going to write code in the AI first world I think there is a reinvention and I would say uh developers and data scientists I think leveraging both your strengths right I think to hands if you can raise your hands uh what about data scientists yeah I definitely see that so interestingly this world is going to massively shift the way developers think about their jobs we heard about the massive productivity increases uh in one study we found out that um with AI pair program you are nearly 55% faster to kind of spit out a uh a jscript kind of application so sandip I I really think at the core of this is that productivity gain right for developers it would be about completing code about unit testing Automation and we seeing massive shifts at Microsoft my own team is beginning to adopt this at a pretty rapid scale and we're seeing that massive productivity gain but for the data centers out there uh the models that you create the way you're able to build those models and test those models and kind of put them out there I really think there are massive again platform capabilities say the Azure AI Studio as an example you can uh there's models as a service so this I think really rich capabilities in the underline platform and when I reflect on like how we used to do work in the API first world versus how you're going to write code in the AI first world I think there is a reinvention and I would say uh developers and data scientists I think leveraging both your strengths right I think to it's really going to be the Nexus of these two communities coming together to create these very intelligent applications so I really want to talk about that and Microsoft has led this AI Innovation with a lot of responsibility and creating a lot of trust with our customers so deeply thinking about your AI models thinking about responsible AI principles how do you build inclusion fairness privacy trust and thinking about all this in a very fundamental way so you don't it's not an afterthought an AI first application cannot be built as an afterthought it has to to be built into the core of how how developers build an application so wonderful thought Shu in fact for security always said shift left um I think happen for AI also going forward uh Rajiv coming to you um we literally swimming in this world of llms right now slm lams and you know what's going to come in the future in fact sat had a wonderful slide where you had all the models including F2 which is the latest one right um do you think um for developers to succeed in the current times there's a need for a unified platform I think that's Central so you're t uh I'll start by saying look last year was the year of AI this is the year where apply AI is going to go at scale and at that scale you are the center of that scale that needs to happen across the world now the question is how do you build those co-pilot is the best way to approach building an AI application now what does it take it it is a massive stack that Microsoft has enabled you right from infrastructure layer to the models to the app layer let me just give you a little bit detail what does it take as you build this application because those are the elements that will help you build the best AI application at the bottom is the infrastructure L SATA talked about we really have soured the best gpus the most gpus from Nvidia AMD we build our own GPU Maya that is the infr macrosoft has the world's largest AI training and influencing platform actually so that's the beginning then on top is the models the models like of course the open AI model then is mistal all of the hugging face model llama you name it we bring it to you as a model as a service so you when you build your application if you want to change the model for cost reason or performance reason from client to the service to the server you can change it without changing the code that's the second layer the third layer is basically the tool chain that you use Microsoft has built the tool chain uh right from GitHub co-pilot that was demoed then we have the co-pilot Studio as your AI Studio you have all the right all the tools to build from low code to no code and build the AI applications but then on top you have to leverage the data and and we have the infrastructure and the and the the OS which is the for doing rag for example vector vector uh Vector search right so that is there and but on top of that is the data that you bring that data is your data but you have to actually we what we provide is the no well first of all data we provide REI we provide you with security compliance uh privacy all those things all built on this compute which is the confidential compute so that right from the compute to your data is ensured that is within your control within your premises right your data is your data we don't use it we don't let anybody else use your data and that is very important so that I think is the Stag that you can use to build world's best AI on aure perfect stop that this is a very interesting initiative launched in India and Asia to really help us bring more uh diversity and inclusion into our communities so code without barriers for instance addresses nearly 75,000 female developers in in in in this in these communities across 50 plus open source communities 30 plus women in Tech communities so we're very happy with that and I think a lot of Skilling uh Skilling to your point right I think we're very excited about that wonderful Rajiv um to is a do moment right so your thoughts on that I know we Runing out time but I'll tell you as a developer I'll get to the chase of it the way you write code is going to be different the way you approach is going to be different it's no longer a linear set of code that you write because remember what SATA talked about this new platform has two most incremental in incredible things one is the way you interact is all natural language it's about voice text video image the second is the reasoning engine so it's not about writing code to get something out but really you have to figure out a way to prompt this this computer which is the reasoning engine to get the right response for to solve your problem so the way you approach is going to change I feel the two things that is going to be true is one every application existing application has to be reimagined on this new platform anything you're doing is are you using AI you should ask that second new application will get built on this platform that was not at all possible ever before okay now in the end because you're developer many people worry about will AI take your job let me tell you the answer to that no they're the pilot developers developers who use AI are going to replace developers who don't use the power of AI yes that's your message awesome awesome Rajiv charu thank you so much thank you so much g once again for Raj thank you wonderful thank you thank you awesome best Miss yes perfect so with that we will we will basically uh dispers from here but before you do that uh quickly are you pum up again in terms of learning learning is primary you all of you here for learning uh if I were you I would do this right there is a connection Hub outside spend some time there understand Solutions people have built how can technology make a difference society and of course choose your breakout slots right end of the day you you you You' basically chosen a stream to walk path in terms of you know what I love AI but I love this part of AI so let choose specific stream so I would urge each of you to look at the agenda once again look at what sessions interest most and spend time there right I think by end of the day today I can assure you that each one of you will walk with immense knowledge right as Microsoft I've seen um these event happened you know year on year but this time I can tell you with AI we have some of the best speakers who there in this venue their intent is to educate all of us yeah with that thank you so much have a wonderful rest of the day and uh have fun programming have fun recing [Applause] bye-bye please enjoy your coffee break at the poolside
