hello everyone welcome to another innovators live and another beautiful cloudy day I'm Debbie Cabrera your host for today and we also have a wonderful guest for today's session where we where we'll cover building applications in the AI era before I introduce my guest let me review a few things here first and foremost our chat if you're joining us live you are able to ask questions and share comments in our chat and one of our moderators will do their best to answer them and we'll even select some to answer live at the end of the show if you're not joining us live the chat feature isn't available but you can see the questions that were asked and answered no matter how you're joining us today make sure you go to cloudon a.wig google.com and you'll see all of our upcoming live episodes and all of our ond demand sessions as well okay let's get on to the exciting stuff Our Guest today is going to take us on a journey through a fun and educational AI presentation and demo bologi rajam welcome bologi it's always a pleasure to see and to work with you how are you doing doing well thanks for having me I think this is my fourth time and I'm supposed to get some gift as part of this yes you are a veteran a veteran innovators live um so thank you so much for being here uh you always have really insightful information so I'm not just saying this I do love having you on here so can you give us the scoop on what you're going to share with us today yeah we will uh um talk through what are some of the things which we are hearing from our customers and uh we will talk about how things have changed since like 2023 which was the year of massive gen hype but now that is starting to land and people are starting to really look at how can I use this to improve my business value uh to my end customers so we will go through some of the key points which we are hearing from customers and then we will do a fun demo where we build a recipe app uh using our gen tools which it's not meant to be a very complex app but it really illustrates what are the tools and services which Google cloud has for you to quickly iterate and build a pretty decent app yeah and I love this demo we've talked about this we love food right we we're both big food fans so this demo is actually really helpful because it helps you look through all the recipes that you might have uh lying around yeah and if you have any questions about the recipe there's a chat bot where you can ask questions and we will show all of that later today and uh for people who have registered ahead of time you probably will have access to a quick laabs where you can it has a it is a built-in environment where you can try out this entire lab start to finish without having to set up any of your environment so uh please check that out and for people who are joining us later I think we will have a Cod laabs Link in the resources tab where you can go and try out this lab there are detailed instructions for you to try out yes exactly and we'll go through how to access that quick laab in a slide so don't worry about that uh but for now do you want to get us started let's get started so these were typical yesterday's applications say you are a family traveling to Greece and you're looking for kid-friendly hotels in Greece you go to your typical website and search for it and it throws up a whole bunch of results and you are left to figure out what exactly is a best match for you so but model applications a lot more interactive right it asks you questions tries to understand what your need is and really gives you the results which you're looking for so that you're not uh left with the task of figuring out things for yourself right it is much more intuitive and personalized and that's the direction which we are seeing and even now uh the expectations of consumers have changed where we if we go to a chat website like a e-commerce site we expect be able to ask questions and find out more information so when you to start building those modern applications these require an extensive array of capabilities to be successful first is scale previously your applications were restricted uh most of the time catering to like a very small audience which is based in your geography and so on but that has changed right but with the power of social media and so on you are able to reach consumers far from across the globe so you application needs to be able to scale to serve millions of users so the other one is uh security right which is becoming very critical because there is always some amount of uh security threats going on at any point in time so your applications you build have to be reliable secure and continuously available that's also a huge thing because if you are running an e-commerce site you cannot afford to have like a 8 hour down downtime right uh one of the customers whom I have worked with in the past said one hour of downtime cost them a million dollars they are a fortune 10 customer but it may not be that as big for other customers but it's still important and then you need to have the ability to innovate right provide new and U Better customer experiences to your end consumers so that you can differentiate yourself from your competition uh so this uh the platform which you use needs needs to give you the ability to experiment quickly and move faster and ultimately lot of this is built on data foundations where you have to have ai ready databases that you can use to actually power your application because gone on the days of using Legacy databases like to power your applications you need databases which can do a pleora of things so like when you start building applications like that in the past that's not going to work in the modern era and when people who have worked in like traditional Enterprise applications you have the vpf apps and you have the vpf databases so the vpf apps focuses on developer toil agility cloud observability and security and the VP databases focuses on database performance scalability uh security of the data U making sure their data platform is future proof and you have advanced capabilities for building gen applications previously they used to a lot of them used to operate in silos so there used to be this whole um U dance which needs to happen to coordinate with the database team to ensure that you can deliver an application on time and within budget but now they need to be lot more integrated together so that they can work off of each other to build the kind of services which the business needs yeah and just on that I feel like I say this a lot maybe if you heard me before I'll sound like a broken record but yeah data and ai go hand in hand right you're you might have great AI but if you don't have a database that can power it if you don't have good data it's not going to be as great as it can be really yeah and um ultimately when you Leverage The Power of AI the data is your secret Source because you if you have been a business who have been around for a long time you have so much data which you can leverage to personalize the AI Services people who have used gen tools know that yes gen tools provide some certain answers but when you personalize that with your own data and ground that in your own data that's when the true power of AI is unlocked yeah absolutely cool so when it comes to um building modern applications if you look at uh the operational complexity which uh Enterprises deal with that is is actually hindering the in ability to innovate and compete there is so much uh time wasted between Contex switching when it comes to uh developers they are in their documentation they are in their IDE and so on so that limits the uh productivity gains which you can have and uh when you look at um um the duplicated effort to build and deploy applications which happens across lot of industries that means that you have a slower time to Market and then you have technical debt and Legacy databases which really hinder your ability to innovate and build more modern Services uh if you're using a very Legacy database you can it cannot be used as a vector store uh for your AI applications that makes it more costly to maintain and it really limits Innovation uh and then lack of standardization also means that it your applications could run into high security and compliance risk as well especially in regulated Industries like financial services and Healthcare this becomes a huge challenge so we want to make sure that the technical teams have the best experience to build a future basically adopting AI supported development practices uh for building your applications so Google cloud is your partner for building modern applications we have infrastructure which is built for scale right our Google kubernetes engine or gke can uh operate at 10x the larger scale than any of our competitors right um and when you look at AI expertise and Leadership ship this whole wave of J was triggered by the Transformer paper which we published in 2017 and and we have been using AI for such a long time it almost becomes kind of second nature to I think we published a paper in 2016 where we talked about how we reduced our power consumption and our data center by 50% using AI right and people people one of the things which people don't see is how we use AI to actually filter out all the spam which you used to get in Gmail I think it is like a 3 billion pieces of spam messages which we filter out which you don't even ever see so that's how powerful and our use of AI has been for the longest time within Google and we also have industry-leading databases where you can have a globally available uh database like spanner which is consistent across the globe uh and available all the time right and we also have alloy DB which is the Improvement upon the postris engine which can function both as your transactional data store your vector data store and and it can even do some amount of analytics and we have worldclass security right since 2020 Google cloud has had 75% fewer uh critical and high city security issues than our competitors yeah few things on here it says uh you know Google uses AI to write more than a quarter of its new code yes this this strikes me because recently I I haven't been able to get my own statistic yet but I feel like it's been helping me write more than a quarter of my own new code as well yep have you experienced that oh uh very much so because uh one thing which I tell people always is when back in the day when I learned computer science I spent a lot of time learning syntax and semantics I think that is actually going away you no longer have to worry about oh is there a colon here or is there a semicolon here because that was a language semantics now you can ask code assist for help and it actually completes that for you so that is essentially getting eliminated and it lets you focus on the more creative aspect of coding yeah and you mentioned alloy DB we recently did a road show in Q4 and we did a few alloy DB Labs those are also available on code labs and I just want to if you want to play around with that you can really see how you can do kind of everything in one place absolutely um so one of the areas which has been a lot of interest to customers because when you start building applications for the more modern era you need to have a platform which can actually cater to all the needs of your applications and when you are a large Enterprise you need a platform like that uh so that uh different teams within your organizations are have consistent way of building applications that makes it more maintainable ensures it follows all the security and compliance policies of your organization so and so on so we work very closely with saber uh saber if you make any kind of Airline booking I think it pract practically goes through saber I think 99% of the time um so we worked with saber to actually help them build out their application platform and we have had a very strong opinion as far as when it comes to platform engineering is concerned and uh another example is a customer called regology where we work with uh them to actually help them adopt alloy DB actually just like what you said uh because they felt like that is a much better uh way to leverage their data to uh uh improve their chatbot and uh provide um services to their customers and the next one was we helped uh we were working closely with tuning and there have been an early partner for code assist right from uh it launched nearly I think well over a year ago uh at this point uh and they saw uh productivity gains of around 33% uh when they started using uh um um code assist and that means that they have boosted developer productivity faster time to Market and much more streamlined on boarding yeah again here with the productivity gains I just feel that a lot in my own experience I was just uh creating an app last week and I had set the whole day to be able to do it and I actually had some free time after that's because you're smarter no it's because of the Cod assist yeah truly it's it's really changing the productivity gains game true and millions of developers are building with Gemini right Gemini is providing so vital wide variety of services uh to actually build new applications especially with Gemini code assist we have seen so much productivity impact internally there's a 40% faster completion time for a common uh Dev task 50% less time writing tests 55% less time writing new code because the code is actually getting generated for you and I think one of my favorite things is actually the 55% less time understanding new code AS developers you you are always thrown into fixing or updating somebody else's code and uh back in the day when I used to be a developer um writing code every day used to spend so much time trying to do a dry run of the code trying to understand what exactly a block of code is doing and now you can just write select the code right click on it and say hey Gemini explain the code to me yeah and also another one that I love is that you can use Gemini to comment the code as well so not only do you make it easier for your yourself but also for that other person you have the comments in there and you have the explain this function exactly for a lazy dollar person like me who don't write enough comments that is a fantastic feature with that I'll hand it over to Gemini uh to uh Debbie to talk not Gemini Debbie to talk about what's next you can call me Gemini if you want no no kidding kidding of course okay so what's next and that means what's next after this event first you can continue exploring this lab so once this live broadcast concludes today you'll have the opportunity to explore the lab in another platform as we uh bology mentioned before Cod labs and you'll receive instructions on how to do that in a follow-up email so don't worry if you aren't able to do it now or you haven't registered early what else you can try out more Google tools so you can join our innovators community and develop new skills with 35 no cost monthly credits to use on courses and Hands-On labs in Google Cloud skills boost which is the lab that we'll be uh using today in the live demo or will be the lab that you get access to today and then number three join us at Google Cloud next 2025 in Las vas discover new and upcoming Innovations at Google Cloud next and you can get inspired by the technical talks we have you can tackle School skill boosting challenges and get your questions answered as well as connect with other fellow developers and we're really excited to offer you the lowest ticket price at $999 if you register by February 28th using the code Dev 999 VIP at checkout to redeem this offer I'm really excited about next this year very much so do you remember the penalty kick demo which was there on the floor last year yes which was fascinating to me that was a cool use of AI uh tools yeah yeah and I know some I know some other people that are building some cool demos uh for this year as well so yeah and the exciting thing about coming to next is hearing from all the other people who are using Google cloud in so many innovative ways and you hear about use cases and scenarios which you normally wouldn't 100% getting in front of people that you don't normally see every day it really is the best Okay so let's get into accessing the lab so what I like about this lab is it covers a lot of the points that you made right bology about modern data development data foundations and how you can Infuse AI into that and kind of make them go hand inand um and that it uses a lot of different products right we'll use a Vertex a agent builder of course Gemini code and Cloud assist uh big query cloud storage collab Enterprise notebooks and more so we're going to get a good view of all the different products many different products and services that you can use in the cloud uh I really love this demo okay but before we jump in let's go over how to access the lab so if you're joining us live live go ahead and open up an incognito or Anonymous window in your browser and then you can go to the quick laabs URL we have on the screen here explore. quickls dcom once you're there you can sign in with your account now make sure it's the same account that you use to register for this live webinar your email was previously registered so if you try to sign in with a different email it will not work just make sure you're using that same email and if you're just joining us and you didn't register earlier you may not have access to this lab but the good news is there will be a very similar version of this lab as we mentioned that you can complete in COD Labs on demand once once this event is over and again you'll receive that information via email so once you've signed in you can go ahead and click the Box building apps in the AI era technical demo it'll look like this image here and from there you'll be taken to the classroom to start the lab now very important don't click Start the lab until you're ready to start the lab because there is a time limit on these labs and once the time is up you will not be able to pick up where you left off you have to start from the beginning these do have to be done in one shot the good news is that this version of the lab will be available until the end of the California day which is where we are today 11:59 p.m. Pacific Standard time so you have all morning afternoon or evening depending on where you're joining from to complete that y yeah so if you want to do the lab after today as I mentioned there's that on demand version and you'll get all those instructions okay and if you're having any trouble reach out via chat to see if we can help you get access all right let's go on to the demo thanks debie and uh at this point uh um one thing which I want to rrate is two things one is um making sure you are in an incognitive window we use Quick laabs all the time at our various events and that inevitably happens where people open it in a window where they're already signed on to with a different set of credentials and it creates issues and the other thing which I want to say is I think the lab is for 4 hours or yes 4 hours but um we did it in events and I think most people were able to complete it in an hour or an hour and a half I think yeah maybe yeah it probably won't take longer than two hours uh but just in case there has been times where I have been like oh I'll be right back and then we yeah don't start it when you're in the middle of a three-hour meeting yeah yes cool so the first thing which we want to show is around Cloud workstations so Cloud workstations is a environment in the cloud I love this so very much because it lets me have different workstations for different demos and stuff which I do so I have one for this I have one for the different set of demos I do so I don't have to Tinker with the environments each time I need to do a demo and the nice thing about this is uh it work you can actually use it with your um local ID as well you can connect from your local ID into your Cloud workstation and use it as if the entire thing is on your local machine but from a organization perspective it makes sure that all the files and the data stays in the cloud right so once you here uh uh the pre-warming step of the quick laabs should already have created the cluster and have the uh workstation ready for you all you do is if it's not started click Start it'll take couple of minutes to start up and then you click launch and then you will get a window like this which kind of looks like um the opening of it's actually a managed vs code right vs code instance and you will download the code to your machine and uh this is actually what it looks like so it's a very simple application right it's a recipe uh search and a cooking advice application fun fun stuff um and uh we will go into the uh uh code a little bit later but one of the first things which you do as part of this is actually creating um a data store for an agent agents are all the rage these days uh so we will actually start creating an agent so we have something called an agent Builder and so obviously since this is a demo I have done all of those steps yes just like the cooking shows right uh so we have data stores here in one of the data stores we are actually what we are doing is we actually just storing files so if you go to recipe uh old cookbooks folder and if you click on one of these files this is actually The Complete Book of cheese oh wow my favorite book yes uh Complete Book of cheese and we have a whole bunch of um cookbooks which are in the public domain which we downloaded and uploaded to a Google storage bucket and these are old cookbooks right if I'm not mistaken yes yeah these are all in the public domain so we don't want to viate any copyright so absolutely yeah so we uploaded all of this to our Google storage bucket and here uh when you do create an agent you can actually specify a data store which can be like a bigquery database um a cloud SQL database it can be a storage bucket as well so what happens when we do this is it actually takes all this data and it's almost creating like a inbuilt rag like a retrieval augmented generation so it indexes all that data and does it for you so that way you don't have to deal with the task of oh I need to find the right algorithm to do the in the create the embeddings I need to find the right um algorithm to find the uh do the searching and so on so all of this is taken care of you taken care for you because that makes it much more easier for you to build an application so once the data store is created you actually go back and you can actually create an app so in this particular case we have done a recipe search app which actually uh sorry this is actually the The cookbook search app app where you can actually go and search and the fun thing is you can actually search and test it here too before you even build app but we have actually built an app which we have deployed on cloud run already so in this case we can actually search recipe chili recipe oh chili delicious it's cold so I thought we might use chili so one of the things which we have done in this particular search is we have actually uh grounded it using Google search as well so it is actually pulling stuff from um the both the uh books which we have have and also other places as well so if you look at these are all the results which it found in the QuickBooks which we uploaded so it has things mothers used to make a collection of oldtime recipes right so so this is a quick way to create a agent and to actually test it out as well once you uh done with that you can actually go in and create another data store I think that will be one of the steps in your uh this where you actually create a data store against a bigquery one so for the bigquery one one of the things which we are doing is we have a spreadsheet uh uh with lot of uh information about recipes and that we are going to use our collab notebook to actually upload to bigquery and this is actually one of my I am not a very smart person when it comes to uh collab uh I am pretty dumb I know how to be dangerous you're inexperienced let's say that yes maybe that maybe that you you put it much more nicer than I did right so what it does is it actually taking a CSV file which has all the information and you can actually use um uh uh collab Enterprise to actually consume that data and upload it to bigquery so what we are doing is we are taking the uh information parsing it retrieving the right information and pushing it to bigquery so one of the things which it can do is it can actually suggest code as well so actually you start doing it as you can see it types ahead and gives you suggested code so so code assist actually works within your collab Enterprise as well so if you're somebody who is new to collab Enterprise the easy way for you to get started is just write a comment in natural language and collab Enterprise will generate the code for you yeah and I love the you know doing it this way because you said like you can play around with it first you can test different things you can use Gemini cusis to adjust the code if you want you don't have to go and change it in your app right away you can kind of do it in here and see how it works you can ask it in natural language to remove comments uh sorry remove columns rename those columns add things so in the lab you'll see that as well you kind of test out all the different things that it can help you with along the way um and make this even faster yeah and it has an underlying uh runtime engine which can actually run this code and give you live results as well so once you do that you have pushed all the data into um the the um the bigquery instance which we are using and that is actually what we going to connect uh using agent Builder into um into the uh data store yeah and I will say I've used agent Builder uh a few different times at this point probably like seven to 10 different times and not only is it constantly improving I feel like each time I go in there I see something new but it is really quick like in a couple clicks you do have a search app or a chatbot or something like that spun up to kind of try out things it's very very fast yeah the first time I tried it I'm like I I don't understand anything about gen let me try what this thing is and I was was very surprised with the results which it was able to get yeah um so uh this is the end result of actually processing the uh spreadsheet data using um collab Enterprise and uploading it to um bigquery so you see all the data which has been upload loaded already this is a schema for the database table in bigquery where we have the directions the title and the URL and you see all the individual documents here and after that we actually built an application one was for uh searching through the cookbooks which we showed you earlier and same thing for a recipe search as well so um both those applications have been created now uh let's go to the actual code actually uh it's a very simple uh streamlet application um somebody I don't know much about streamlet somebody showed it to me it was a very simple way to create a quick web application so it was very helpful yeah I actually used it that app I was telling you about that I was creating last week I needed to put a quick front end on it so easy and simple just pull it in make a few edits it really is great yeah so this is actually uh some of the code which you see let's ask what is this code doing so I'm going to use Gemini CODIS to say explain this to me so let me in this part oh yes this is a great thing to point out since this is a separate uh area like if you're using a Gemini Cod assist anywhere else you might have to log in here and just make sure you're in the right project maybe enable an API uh because we want to make sure that the right project is logged in for COD assist as well as for whatever else you're working on so you will find this uh every time I'm doing a lab or something I find myself yes you have to log in a few times just to make sure that you can use it in the right place cool and so now Gemini code assist has gone ahead and explained the code to me so it's saying initializes a chat session with a large language model and configures to use specific data store for grounding its responses and it's also breaking down the individual steps of what it is doing so it's a very easy way for you to understand what the code is doing and make changes to it much more easily uh So within your code please try this out I know it is tempting to just copy paste the commands and just rush through the Cod Labs you have four hours to do the code Labs so please make sure you take the time to try out all the things which we are mentioning in the quick laabs as well as the code Labs uh so that you can actually understand the power of these tools yeah especially if you're used to programming in maybe a different language than what some of these uh Labs have or you know what this app is in uh then you can really understand what's happening by getting the full explanation and it's very thorough as he said it goes through every single line and tells you what's going on exactly and one of the other things which I used to like to do in code assist is uh can you create a test plan for this because I used to write hate writing test plans oh yes testing very important very important yeah of course we all do it yes must we don't we don't want to test information yeah some people do it but I prefer not to so it's actually going to create a test plan and give you instructions on what should be covered as part of the test and um uh let's see so uh what it's actually using um unit test mo. patch to MOX or Vex AI in it uh and it
is actually providing you all the steps which is necessary to create a test plan for this right um you don't even have to leave the ID right so you have your code there you're like oh let me I need some test can you just help me with that and also if you have any other questions about the other files that you have open or other Google Cloud questions you can ask how to do it right here in the Gemini cusis chat in the ID I think that's a great point because very often all of us used to have like stack Overflow open Google Search open documentation open while we writing code this actually lets you stay just in this window and ask the appropriate questions to move ahead with that yeah you know we all have well maybe not all of us us some of us have many many tabs open all the time and once you go out into tabl land it's hard to come back sometimes how are you doing on time jimy I think we're good on time okay yeah we have 30 more minutes wonderful so let's look at U the the code where we actually calling the uh uh the app so in this particular case we are actually making a call to Gemini to actually retrieve the appropriate data when uh there's a chat message is uh sent and let's look at the other recipe search one which is also doing something similar where we actually going to um the let's actually do an explain this on this so live demo I love trying out different things I feel like he asked me how we were on time so he could show this special piece here which is exciting explain this file and I'm not joking when I say that I use this all the time I know it sounds corny like oh they're paying her to say this no no really I do I use it all the time it's just very helpful because even working at Google Cloud I mean you've been here what 10 years eight years eight years yeah my me five even all that time there's so many products so many services so much different knowledge that it's nice to have that friend there that you can kind of just ask different questions I know bology and I both tend to work late at night you know so when you can't uh you know ask other people those questions because they're sleeping Gemini Cod assist is there to help you if then you can't phone a friend you ask Gemini right yeah so uh so I did uh explain this file in this and it's Gemini has come out with explanation of the entire file on all those steps which is doing the import steps the search sample step where it is actually uh retrieving information from our data store and it's also telling you how the streamlit app is functioning right whether you when you the query block where it is actually get retrieving the results and displaying it back to you and we actually did deployed this on um Google Cloud run uh so let's go to um the the agent builder stuff where we uh actually we go let's go to Cloud run yeah where we actually have this application deployed so Cloud run uh have you used Cloud run before Deb yes I use it a lot I love deploying my apps there I feel like it's the easy fast way to deploy exactly because it is completely serverless you don't need to create a cluster or really understand all the tuning which you need to do to make an application function well so you can easily deploy it and the nice thing is it only really consumes resources when there are actually requests coming into it right so this is a recipe app which we have deployed and let's go into that app and here's a URL for the app which has been deployed and and what you mentioned Cloud run right so when you're not using it you don't have to worry about undeploying it or anything like that it only uses resources when you're you're calling it when you're opening the service got it so we have our recipe search app in here and we have our cooking advice app in here so let me ask the quicking advice app on uh for some recommendations uh while that's coming up oh yeah I wonder what what should I have for dinner today this evening cooking advice I'm going to ask like I have some be friends who are vegan oh okay so let's see what can I use as substitute only if I can type better maybe J I can help with that actually a quick note on that sometimes if you misspell things uh not in your code but if you misspell things when reaching to your AI uh it'll actually still understand you which is very nice maybe you've used other llms and chat Bots that are out there and you notice this but it really makes it so that in natural language you know not all the time do we type correctly or say things in a way that is completely comprehensible you know so so I'm going to pretend I made that spelling mistake on purpose oh pretend it was totally not so I'm asking hey what can I use as a substitute for butter to make a recipe vegan so it gave me a whole bunch of options like applesauce uh neutral oils coconut oil peanut butter almond butter all those are options oh that's great and and honestly I have a lot of dairyfree uh friends so you know making brownies or anything like that applesauce I hadn't considered that fantastic nice cool let's look at recipe search uh let's say I have a lot of mushrooms in my house mushrooms yeah not not the magic kind mushrooms just right um mushroom recipes oh mushroom migas I've never heard of a mushroom migas me neither I love migas though well this is so this is great because you know when I want to make dinner and I say okay I have one leak and some turkey you know what can I make with this um this is wonderful right and we created this in what I mean in total it'll be like an hour an hour and a half at most yeah and you just have an app that can help you with everyday tasks of course but also in your business with other with other tasks aside from everyday uh home tasks fantastic so that pretty much concludes our demo awesome yeah and great demo yeah nice that's uh is there anything else which we should show well no I think actually we're going to go to some questions but I do I just want to say that I love this demo uh I mentioned because it touches so many different Services right we went to agent Builder we had big query we created the data store uh then we used Cloud run and it was so quick and it shows kind of like the sheer difference between how much faster we can develop applications and software now because of these AI resources that we have and how seamlessly they integrate into our current development life cycle right we've both talked about how much that's changed for us uh so now we're going to take a quick break for questions and we'll select a few to answer live on air and we'll see you in a sec and we are back okay let's get into answering these questions let's get right into it so a question came in which tools to choose When developing an AI app it the the broad answer is it depends right It ultimately uh depends on the use case which you're trying to service because with developing there's no AI app per se right AI app for me means like the Gemini app right more about like apps which are powered using AI I think that's the way I think about it I think that's probably what the person who asking the question is intending so depends on the use case whether if you're using say U the great example is what we showed in the demo today and which you can try out on the quick laabs where how quickly you're able to leverage AI within an app right uh because the if you look at the overall code which we wrote the part about utilizing AI is probably like two or three lines at most rest of us was a standard app all we did was made the call to the right API and retriev the right information I think that's where it boils down to and if you are looking at more complex applications where you need to chain results from one Ai call to another and make decisions based on that then you start looking at tools like Lang chain and whatnot yeah and you know to go along with that you can decide on do you want to Lo use a proprietary model an open source model an open model there are many different uh you know ways that you can access an llm as well so and I I think I think that's a very important Point Debbie because one of the things which verx AI lets you do is lets you choose your own model because we have our foundational models which are Gemini I think the Gemini 2.0 flash which went GA last week I think and uh we also have like third party models like models from anthropic Mistral and so on available and we also have a whole variety of models from hugging phas which are open source models available for people to use so pick the best model which is ideal for your use case I think that'll be my advice yeah exactly and even uh the most recent deep seek you can actually try it out on vertex AI oh I haven't tried that out that's be fun and I think vertex has over 300 different models you can try so yeah perfect okay next question so why is AI important for modern apps um more than AI important for modern apps the way I look at it is customer expectation has changed so drastically right I think when we were all stuck at home for Co our expectations of what an app should do or what uh the kind of customer experience we hope to give have hope to get has changed quite a bit um and I think one of the stats which I read a long time ago is it takes you 66 days of doing the same thing over and over again to actually become a lifelong habit and we were a lot of us were stuck at home for a lot longer than that and that has really changed our expectations out of how we look at um uh what we need from a company or an Enterprise and so some of those functionality for instance being able to have an interactive conversation about a particular product say if you're buying say even a prepared food from a store you want to be able to ask hey what are the allergens in it say if you're allergic to certain things you need to be able to really understand how uh it works and things like say if you're buying a piece of clothing hey everybody has a different body type how would it look on a certain body type being able to ask those questions and interact with it or maybe even see that piece of clothing on different body types I think that is a feat one of the capabilities which we released a long time ago um as Google and so those kind of capabilities are becoming more and more important into the app and I think that is why providing a more modern experience is made easier using a I think that's a way I think about it yeah and and also what you were talking about before right with the personalization so the body types or you know we've the the sunglasses try on things right every day that's getting better and I would also encourage what you mentioned before like look at your use case don't just assume like oh I need to use AI you know I there's also an argument to say like AI can probably be integrated into almost any use case that you have but really don't just try and use AI because you think you need to really take a look at your use case and understand how it would benefit you good okay so next question so how would you integrate AI into an app there there are a bunch of different ways right we mentioned some of them but yeah you can do what we did today you can make an agent you can call a model into your app to uh bring in grounding from search or rag uh you can use it to structure your data or collect your data recently uh with a group we were doing a little bit of a hackathon and we built an app where we uploaded pictures of our clothing and then we used the model to kind of put metadata to it and say like this is a black blazer this is a sequined silver Blazer so kind of playing around with the multimodality of it too I think can be really fun and you can find a lot of different ways to integrate uh in that way yeah and again goes back to the initial point you were making right around what is the use case which you're trying to solve for and that also dictates to some extent how you integrate AI into your app I think I think uh you were there last week at developer week uh here in Santa Clara where one of the demos one of our colleagues did was he was running everything locally from the browser he was using web Ai and uh G JMA JMA 2.0 to actually power an application which was completely running locally with no internet connection because he wanted to show that as a use case and that that and that worked really well so It ultimately boils down to what are you trying to accomplish as far as a use case is concerned and I think that will determine your strategy yeah absolutely um oh this is a good one alloy alloy DB I feel like that's really your wheelhouse uh are there any worthwhile performance benefits of using alloy DB in a non-generative AI app versus just staying with something like Cloud SQL uh uh definitely I think one of the even though the alloy DB uh product is built on top of a postgress engine we have done so much work into improving the performance of it uh I forget the exact stat but it is so much more performant than the standard Pro PR uh instance which you get through Cloud SQL so and the other thing is alloy DB also provides you the capability to do analytics so if uh if the use case for your um application is hey I want to I don't want to invest in a huge uh data warehouse for my application but there are certain analytics use cases which I would still like to use I think alloy DB is a fantastic fit for that yeah absolutely okay so how do you collect data for AI models so again there's many different ways you can do this there are you know obviously if you have your own data that could be csvs PDFs uh uh you know Word documents anything like that uh or even now going into the multimodality right videos images different things like that and uh then also you can it it can help you structure your data so even if you have unstructured data uh it can help you do that as well as we have all of our public data sets that you can access right and uh kaggle is a fantastic source for data sets even before gen I have downloaded different data sets from kagle to run models against it uh I think once I downloaded a whole bunch of x-ray images and used our pre uh tool called autl uh to actually see hey what can it detect what's happening in a particular x-ray so kagle has amazing data sources uh uh data sets which you can use gets uh started on it yeah and we also have a lot of public data sets in our bigquery bigy bigquery has a lot of great public data sets Okay so what uh what is the list of programming languages that it supports I'm going to assume we're talking about cist here yeah so I think there's over 20 different languages that it supports yep right so the fun part the fun part about the answering this question is there are 20 plus languages it officially supports right um I remember I think it was last year's next when somebody came up to me and asked hey does it support power Builder I said I've never used powerb Builder let's try and I gave it a natural language prompt and it generated a pretty good power Builder script and it was not according to me that it was good because I don't know anything about power Builder they said it is a pretty good script so the the thing about the llms are they trained on so much data I think the languages which we say we officially support are the ones we are continuously regression testing on and we have done enough validation on but a lot of the time I've seen it supports a whole lot more than that yeah and just to we did get an answer in the chat for exactly the languages so I'm just going to read that out but it is there if you want to look at it bash C C++ C Dart go Google SQL Java JavaScript cotlin Lua mat lab and I'm going to stop there but you can keep reading there's a lot that you can play around with and like bologi said maybe play around with something it says it might not support and test it out kind of push limits there okay if I have little to no experience coding can I use these tool to build a complex application that cages to millions of users or will I need a team of developers also if I already have the uiux done can I upload to the databases files and the AI will help generate the back end so um I remember one of our colleagues had a like a graph of the benefits of coding assistance uh so the benefit of coding assistance was one the why AIS and the experience of the developer was on the x-axis so the mapping was people with like little to know experience coding assistant are not that helpful if you have a reasonable amount of knowledge and then at the other end of the spectrum people who are like expert coders they're also coding assistants probably more get in the way rather than actually helping them uh so what we have seen is yes if you have a little bit more experience with coding that is much more easier and coding assistance can generate code for you but it is not a substitute for understanding how to design good code how to structure different code and so on so it understands it at the code level but it doesn't understand it at overall application Level the one of the questions was hey if I have the database and the back end already built can we build that application right actually one of the things which I've seen is people take the schema and convert it to an AP I right um um I think one of the capabilities within um Gemini code assist is the ability to generate open API spec 3.0 right uh so it can you can actually give probably give it a database schema and get it to generate a spec and from there you could probably say hey from this API spec generate a stub app for it obviously it'll just be like a marked up app you'll have to connect it to the back end and do all those things but it is quite possible and we have tried some of that out as well yeah and I will say to reiterate what you were saying the human in the loop portion is something that we you know constantly remind people of at events and webinars like these because um it's meant to help you Gemini cusis is there as a helper it's not meant to do everything for you right so you will have to go in there and check and sometimes you will have to make edits because you want to do something slightly different than what it gave you so yeah make sure you're kind of checking in uh I don't know if it would uh yeah make sense to cater to millions of users just off of what it gives you especially not in production so always make sure you you check that y uh okay so how much does AI app development cost that is a broad question so if you're talking about um creating a very simple app using Gemini APS it doesn't cost that much it is fairly straightforward to do but if you're talking about hey I need I want to trade my own model then costs start adding up so I think I don't know if there's a better way to answer this question than that yeah I would agree with that and it also yeah it depends like how often are you going to be calling the model right like which model are you using there's actually pricing online that I don't know off the top of my head but there is pricing for how much each call input and output is for our newer models you know I I know definitely 1.5 and 2.0 are on there um so I would encourage you to go check that out yep we have a handy pricing calculator on the web to actually try that out yeah exactly okay I believe that is all of our questions for today yes yeah thank you so much for engagement yeah yeah uh we love to hear your questions so you know thank you for being here for engaging uh any parting thoughts apology um no the only thing which I would say is that quick laabs is a lot of fun to do yeah uh if you already have access to the quick laabs try it and if you um are joining us later on demand try out the Cod laabs yeah and if actually I'll answer one thing here maybe some of you are saying quick laabs code Labs what's the difference here they're just two different platforms where you can do the labs quick Labs kind of sets up the environment for you you can get right in and start doing it whereas the code lab you uh need to do a little bit more of that leg work right so Cod Labs will be available indefinitely and there will be some free credits that you can do and you'll receive that via email and the quick lab as we mentioned will be available until 11:59 PST so uh please please enjoy doing that thank you so much for joining us for another innovators live uh make sure to go to our website and check out our past sessions and our upcoming sessions and let us know on the web if you complete these Labs you know baly and I would love to hear from you how you feel about it thank you so much
2025-02-27 04:03