thank you and uh welcome everybody today to the Deloitte AI Institute Canada's session on generative AI in the consumer and Technology media and Telecommunications industry my name is Peter Danforth I'm the partner in Deloitte that Services AI for Consumer and the TNT industry before we begin I'd like to start with a land acknowledgment we acknowledge the indigenous peoples of the lands that we are on today Deloitte Canada has offices with representation across most of the country we acknowledge that our offices reside on the traditional treaty and unseated territories as part of Turtle Island and is home to many First Nations metis and Inuit peoples lastly I'd like to remind you that this session will be recorded and available on the AI institute's webpage and will be available in the language of your choosing so as it comes to this topic there's no denying that we live in the AI era now um over 50 percent of c-suite Executives that were surveyed believe that AI can improve their customer care services and 47 of c-suite Executives believe that AI will transform the way they effectively manage cost and inventory going forward as we think about how these new technologies and the adoption of them can change the way we do business at scale and quality there are many challenges moving from technical concept uh into actual scaled execution those include large data Legacy systems analytical platforms decentralized data and lack of centralized Authority and responsibility for these tools as we scale these Technologies it also creates new problems for us in terms of how do we trust Ai and what should we be doing and allowing it to do from an ethical and privacy perspective these challenges cannot be separated from each other as you think about deploying AI at scale so as you work through understanding The Virtuous cycle of people process and Technology as it comes to applying these applications it's important that we look ahead and consider how these Technologies may affect the way companies need good services and customer experiences forward some examples of AI had used today in these fields includes optimizing networks for both ground and air fleets maximizing efficiencies in the way that goods and services are delivered to customers and to locations personalization is a large area of concern going for AI as well how do you consult a platform's visibility to customer interactions and use natural language processing to automate some interactions and make them more effective for customers both in terms of experience as well as speed disservice as we think about generative Ai and how it might impact our Industries there are two big examples that come to mind the first being you can use generative AI to create digital Twins and those digital twins can be used to simulate all sorts of customer experiences flows through airports how customers might interact with a certain shelf or product the second is in technology generative AI has the power to generate code itself from scratch how will being able to generate multiple iterations of customer experiences or digital ecosystems affect the way we prototype and launch product at scale as we think about these cool new capabilities that have emerged almost instantly and almost out of nowhere to the most people how do we make sure that we don't rush into the adoption of them and create risks and privacy concerns and ethical considerations for ourselves so how do we balance this great new opportunity for explosive growth with ensuring that these tools are adopted in a reasonable and ethical way and with that I'd like to transition to a quick poll question for everybody so what ways would you use generative AI in your organization now I know a lot of you are here today to uh get answers to that question so you might not know right away but we'd love to hear from you all so please scan the QR code or enter the digits below the website Mark once you get to the the slido website see chat Bots content email writing market research all great answers customer retention drafting letters multimedia yeah so clearly we've got some people who are starting to wrap their head around what generative AI might mean to their organization and and we have a great panel here today that we'll get to shortly once everyone's had a chance to participate in this we'll be able to talk more about use cases in in consumer and TMT yeah so very interesting a lot of content production um some a bit of coding is starting to emerge now and project management on the technical side but I think a lot of people are thinking about this from the excuse me content production side of the equation which is is a big area but I think as we get more and more into this space we'll realize how profoundly first iterations and first drafts of code um can be produced and how effective they can be as we uh as we start to play around with these tools at scale all right well thank you everybody for participating in the pool it's great to see where your head's at but without further Ado I would like to uh introduce our esteem panelists for today um and thank them all for participating um in this session today so starting with Mohammed ridwan Muhammad uh is a Canadian with deep ties to Eau and has been raised there he's built a diverse career of data engineering and product development which began at Shopify where he worked on data infrastructure as a petabyte scale in as a software engineer he then joined densa a product team and an early AI startup where he developed machine learning infrastructure and orchestrated production and data pipeline testing platforms following dense's acquisition by cash app or now called Square MO continued and the Strategic development team building new initiatives and strategic components to the exact map ecosystem currently Mo has a is a co-founder and chief product officer at Pluto card the number one corporate Garden spend management platform in m-e-n-a which he helps large Finance teams ensure they streamline their entire payables operation well welcome and and thank you for joining us today thank you very much Peter excited to be here next up we have mercury is the senior manager of PepsiCo who oversees the servicenow digital transformation Journey with 15 years of Industry experience at peptic PepsiCo he's overseen large AI projects that include completing employee experience from hiring to retiring and Digital customer product Journeys with the latest applications in generators AI you can also lead to Citizens kind of developers on Cutting Edge AI summations for sales for Sherman and Fleet teams globally which was our welcome and thank you for your time today thank you Peter good to be here next up uh we have a non-nim car and a non is the head of conversational generative AI for Deloitte Canada he focuses on technologies that deliver natural language processing to customers and employee servicing lines he is it led an engineering team on AI platform cloud and digital transformation programs at some of the largest financial institutions in Canada and primarily focuses on retail Banking and insurance as his Industries but he also helps a great deal and consumer retail as well as TMT so we won't hold his FSI background against him today Anon welcome thanks for having me and then finally Ian Zhao who will be moderating our panel and our discussion today Ian is a senior manager in our Omnia AI group and he has over eight years of experience designing developing and operating Advanced analytics use cases and environments Ian focuses on customer and Marketing Solutions as well as generating value out of data in his journey at Deloitte he's LED Professional Services product management teams and internal marketing operations for various clients as well as at AI startups he built it he built the company where he built a Sandbox and ml capability that was recognized as one of the top one percent in the medium long-term run for by MIT Google and fast company's review Ian welcome sorry I I slipped that up a bit at the end all good glad to be here and have the chance to moderate this awesome panel well with that Ian I'll leave you uh in the good hands of these very smart and fine people and I hope you have a great discussion and I hope everyone participating today enjoys it and gets a lot of value out of it thank you everyone for joining us today great um so with that said um what I would like to do is to introduce the agenda for the day um so we want to answer a few key questions in the next 30 to 40 minutes so that everyone can walk away with a new understanding of generative AI some inspiration on how it can be applied to your day-to-day work and to your organization and last but not least how can you get started so um we're looking forward to uh this conversation because we do have a very diverse um panelist on the team um some of us walk different uh walks um in our career coming from Consulting to technology and coming back to Consulting some of us are in consumers some of us are in technology many of us have international experience um seeing how AI has evolved throughout the next decade right so I'd love to like hear your thoughts on um on your observations so that we can provide some inspiration to everyone who's joining today so with that said let's move into part one um so we've been talking about generative AI but love to spend some time just to make sure we understand what it is so I would love to invite Adnan to provide our Deloitte perspective on what we mean by that um and then we can also give the opportunity to the panelists to talk about how they Define generative AI because it is a big time yeah and so on this panel I think we have a mix of both Technical and you know business focused expertise right so we can cover both sides so with your questions as you uh have them please drop them into the Q a here or in the chat um I'll cover very quickly what generative AI is um you know for many of you who are interested in this webinar this may be um you know common knowledge uh and for those who are new um you know just wanted to introduce the different modalities that exist so uh generative AI is essentially the ability of artificial intelligence to create uh to generate content um often seemingly original content as well across various modalities text images audio code uh voice and video um and in the past we would think of these as skills requiring real intelligence like human intelligence human level intelligence to do so it's a powerful set of Technologies because uh what we're seeing in the in society is that there's a risk and concern that they will be disrupting jobs it will be essentially removing the lowering the barriers to knowledge work as well so there's both benefits and risks um across text generation uh you know there's different applications out there chat gbt being one of them that really evangelize the space um and and subsequently there have been um Investments and new product launches from Google and and Microsoft and um you know text generation is kind of where the field started uh image generation you'll see many examples of that even though I think the title page of this presentation was image generation it's essentially sending in prompts to generate entire images and pictures or art code generation is having the ability to uh while you're programming having a generative AI follow along and suggest uh then you know the next function you should write and it essentially knows and understands your entire code base and adapts to your style video generation we'll start to see a lot more examples of this but essentially from a single prompt being able to generate video uh movies think about even feature like films is kind of the real Target of that um you know imagine watching a movie of yourself as James Bond on the moon um you know you can essentially put in a prompt in every night get your own custom movie and then we'll start to see a lot more audio generation too so this is the ability to generate music that's actually enjoyable and so you can imagine the potential for disruption across the different creative fields and also knowledge work uh next slide please so how does this uh like we'll be covering a few topics here and we may say something so we may say things like a large language model or llm uh we may talk about image generation models or text to image and we'll talk about multimodal models uh language models are effectively um the text generation models that we're talking about so essentially large neural networks that are essentially a form of deep learning where they've been trained to generate text outputs given an input text in a text to image model it's the same thing given an input text or description or an image input it will generate an output image and then multimodal models these are models that are trained to work across modalities so you can provide multiple types of inputs and receive multiple types of outputs or go from one modality to another you know so text in code out or um you know video in video out and it's essentially a single model that can handle it all now how does this really work now when you think about this ice cream if you look at this Iceberg diagram on the on the left um at the core of it is very specialized Computing Hardware uh in many cases this Hardware is only accessible to companies that have made significant investments in technology so typically the the big tech companies out there or the cloud providers it's very expensive to procure or run and maintain and um even requires a significant amount of energy given the size of data that is pumped through to actually train these models the data itself is is collected from all over the Internet so if you can imagine the engineering calendar required to store uh the vast majority of content on the internet um you know that's again reserved for large tech companies the models that we interact with there are architectures that you know go back I'm going to say uh five six years um so some of them are um you know pretty simplistic architectures um but because of the power of the data and the Computing that's that's available um you know you can essentially start to achieve very um compelling outcomes and so for a model the input is a prompt and the outputs are what we see essentially the generations that we see the applications that we're interacting with so there's many applications out there hundreds now that are leveraging uh these models at the core these applications what they do is they essentially compose prompts and com and compose outputs or even chain prompts and outputs to actually create an interface um and so that's the layer that we typically interact with as you know business users or as just the consumers um but it's all powered by a lot of Technology under the hood okay I think I'll hand it off to Ian now I don't believe there's another slide yeah this is great really appreciate it um and I see some um uh folks in the audience are trying to ask questions so feel free to drop any question into the chat box um as we progress um you don't I need to wait until the end um so we want to make this as interactive as possible um but maybe with us like thanks a lot for providing um a quick overview of what we mean by Jennifer generative AI I think two things I picked up on that number one is original content so I would say if we look at the last wave of AI adoption which is back in I would say the most recent one is back in 2014 to 2019 ish with the spark of deep learning techniques and a lot of that are image based use cases but the difference is that we weren't able to like we were able to recognize image better because of the new technique now in the next evolution of now we're able to generate original images and content assets right um so I always call out that could be one of the key differences like what's different now but I would love to pause there and see if Moe and a full star have any thoughts on how we Define genetic Ai and does that make sense to you yeah happy to uh kick it off um I think uh but you guys put it on put it very succinctly um to me the most important part is the uh is the original content um generation part right uh but I think more than just the content generation but to me uh generative AI is a set of like new tools right uh just like how you use your guitar to uh to play a different kind of sound than the keyboard uh the generative AI uh to me is a new set of tools that you can use to create uh various different kind of things whether it's music art um everything else as well right um particularly excited about it in terms of uh personalization which I think Peter touched on at the at the beginning so I'm happy to talk about that at later on in the in the webinar and Ian just to add to you uh more and all the discussion I had there's a huge shift going on um in terms of how we deliver some of the capability uh and the hands of our customer and the hands of our employees even and and this is this is where I see that there's a lot of capabilities coming in there's a lot of it's not about just email the content there's a different capabilities that opening the door and we are just tapping we can't even keep up with some of the stepping that the challenges are coming in and this is where I see that there's lots of rooms there's a lot of opportunities in the in the months ahead or years ahead that's a great Segway into our next topic which is use cases so now with this new capabilities um and some of the challenging the opportunity that we see right in like would you be able to share some of your insights and thinking on um how can you apply this into your day-to-day role as a product manager as a team leader and also as organization delivering value to the customer right um so a lot to like get your thoughts on like the use cases that you're thinking about and then at the end people will also um our team which also share what we're observing in the market so one use case I would say uh that's if the the selling technologies that it's it's evolved right a lot of uh is we had a legacy system pretty much all over and uh very bus at the data driven system that we we have interacts with multiple task points uh with the consumer shift what we're seeing is that there's a new behavior from the consumer and how are we doing it and so one one use case that I could say that uh giving the chat bar in the hands of the the sales people on the front line and based on the behavior which store whatever store you're you're walking in and we can say with the with the 90 accuracy that you have picked up this this product these are the other product you may like and that that is the weak prediction that we can we can do and based on the prediction we can backtrack and we can say hey we're missing self we're missing these are the products on these shelves on this stores at this time and and maybe there's an event going on it's you know you know next door and what is that prediction we're coming in and what what is that we want to uh product all the shelf that we fill in with that with some of the product so uh there's a few other use cases that I can share in uh but this particular one that that I'd like to say that is changing the the front line how the the they behave how they order how they even serve some of the customer and prior to that it was a one-way conversation now I think we we come across from that one-way conversation it is a two-way or or multi-multi conversation even with some of the the consumer uh as we are walking into some of the store as we're we're we're stocking up some of the shelves in in the in the store for storefront got it so it sounds like it's a natural extension to the chatbot experience that we've been thinking about implementing in the last decade or so right because they probably reflect on the Journey of chatbot I think started with a prescripted um interaction say hey like what are you here for pick one of the other three selection that you can have and then you go off from there and I think you know within years I think we got better at that because of um uh the lateral language processing capabilities now we're at the verge of the next Evolution which is now there's a lot more flexibility to how the conversation can evolve and it takes less work for a company to prescribe what the response need to be yep and it's a 30 increase to the bottom line and and and that's that's how it is predicting uh Ian and if if you can increase 30 off your bottom line um with the help of Technology why not 100 and would love to like save um a conversation in terms of quality risks and consideration that we need to put against this so that to Peter's earlier Point as we go into the next wave of genetic AI how do we balance out um the down the potential downside of it right absolutely but we love to like mole I'd love to hear your thoughts on uh what you're seeing what you're thinking about yeah you know um before we jump into considerations I'd actually love to talk a little bit about the other use cases that we were talking about in the consumer than technology one of the really interesting things that um officer had brought up was the use case when it comes to um sales right um I actually had a really really interesting conversation uh with the founder a couple of days back and they've been working on something extremely interesting so as you know in uh you know my company is based out of Middle East uh so the perspective that I come from is Middle Eastern um Eastern Europe East Asia and that kind of region uh Regional world and if you're familiar even briefly with that region you will know that live stream Commerce is actually a very very popular thing uh in that part of the world right um uh so much so that it's it's kind of strange to not see it in in North America but it's extremely popular you know people use Facebook live people who use uh Tick Tock to be able to actually sell products one of the things that they were actually working on was that you know today e-commerce is an omni-channel experience uh a customer is interacting with you from several different touch points they're visiting your website they're visiting your retail stores Etc bringing all of that information together is extremely difficult right so they're working on some new tools where as you are doing the live stream and as you're the salesperson in the live stream based on the panelists or based on the people who are joining uh that live stream uh it gives you contextual things to sell right it's like hey you know uh person a who has joined the live stream has previously interacted with a b and c here are some of the transcripts that you can say live to be able to uh interact with them and convert a high value potential customer into that so that I thought that was a quite fascinating and that I think goes back to the uh the point that mount Azure had had brought up um a couple of the other things that we see uh I'm seeing specific to the region uh one of the really um curious things was that um you know uh this is the first time in history where the Middle East started to introduce corporate tax and and uh nobody is actually familiar with corporate tax in UAE and the surrounding regions and it's a very New Concept to a lot of people right uh so a lot of companies actually accounting firms and others they're signed to experiment with taking the corporate tax manual uh feeding it into uh genitive AI tools so that people can actually ask and query right to be able to actually interact with it and learn uh in deeper because it's just such a huge manual that you cannot fathom and it's such a such a new so it's it's quite interesting how folks are using uh some of the AI tools like that so um yeah so you know I think that's one of the really interesting parts and we are uh as a company as well we are um doing a lot of these things because we serve accountants and few and uh Financial controllers these are the kind of uh businesses uh uh stakeholders that we serve right so a couple of the places that we are using is also around uh just better contextual information gathering right can I ask questions in natural language to be able to generate actual content uh that is useful to me right um one of the other things that we have been uh working on is also um other generation of reports right um you know uh Gone are the days where you have to create pivot tables on Excel anymore right uh you can now if you have some strict structured data which companies like Snowflake and others are helping you provide uh you can actually feed that all into uh models to be able to query them in a more natural way so essentially in that case you can free up capacity of your design team or analytics team to focus on more value-adding activities such as building more beta assets for example oh absolutely absolutely you know um actually uh on that note most of our product managers and product leaders within our companies and also like the companies that we have been interacting with very few of them actually have uh data analytic separate data analytics teams a lot of them are self-serve on their own Now using a combination of various kinds of tools including like uh modern AI tools that are out there um and I I think Ian you had mentioned uh one of the questions uh was around how can you use this to be able to augment your own capabilities as product leaders or product managers Etc maybe I'll just uh touch on a brief point there as to how I personally use it on a day-to-day basis right um so prior to being in products my experience was in engineering and Engineering we have this concept called rubber duck programming right uh I'm not sure if folks have heard of it but robert.programming is basically he placed a rubber duck in front of you and you talk to it about and you explain uh the code uh to the rubber duck and as you're explaining that you tend to debug uh what you're actually trying to solve right uh my version of that is talking to uh generative AI right uh you know I Quarry right I tell it my stories right I tell it how I'm thinking my market research right and as I am doing that as I'm getting a back and forth conversation going right you know you hopefully generate uh newer insights out of that so that has that has actually been probably one of my most productive hacks over the past year right uh very grateful that this is now possible um so that I think I would recommend everyone give better a shot right in rubber duck programming uh the other things that we've been using um AI quite heavily is is uh I run our marketing teams and one of the key things in marketing is that uh you may have a hypothesis on the kind of messaging that might stick with your customer but those hypotheses may not always be true because you're just so in in your business right uh so one of the things that we've been actually able to do is with some newer tools uh we've actually been able to create almost 300 400 different variations of AD creatives right uh which we can automate and be able to run uh as several ad campaigns so that we can see how the how the messaging uh is evolving and what is actually sticking as opposed to us like uh creating those hypotheses so yeah those are some of the some of the things I could go on and on but uh just some of the things that we've been uh we've been focusing on lately that's awesome um and I think that's aligned well with our general observation in the market as well so if we think about the use cases of gen AI if we bucket them into kind of the external customer facing experience versus internal um users right so we touch on a lot of gen AI technology can potentially be used to help doing customer service better because now it's a more natural experience and interaction um so that enhance the customer experience and also potentially decrease the time to issue resolution right um and then there's a huge bucket of use cases around just internal productivity gain and I think we touch on that a few times already and I think a lot of that is driven by this new capability of original content generation right I think more we talked about how your product team and marketing name are using this content so that you can output a lot more in the same amount of time right and I'm also having a very similar conversation with what one of our clients in the grocery space and their marketing team is actually looking into this um one use case is exactly more to your point about how can I now create um creatives um so that I can move faster in my campaign design right in the past um they have about one and a half FTE of content writer and it takes them about one week to two weeks just to write all the content for a campaign and get it approved right now with Gen AI they can output like a thousand variation of that in a matter of maybe half an hour now the interesting thing is that now we're going into kind of like the roles and responsibility changes because of gen AI so the content creator now is actually spending less time making content but more time qaing and selecting the right content from gen AI Solutions right um but just want to tie back to the use cases um yeah so there's a lot of um uh application or people are trying to test how they can use it in the marketing space and especially for this audience who's the audience who's in the consumer slash TMT sector the nature of the the company a lot of the services are provided b2c so um the the volume and the frequency of Direct Customer engagement is relatively higher than other Industries therefore it requires much bigger capacity right now to to help with freeing up capacity augmenting capacity gen AI this is where we see a lot of conversation and testing happening in the Gen AI using the Gen AI Solutions right absolutely absolutely and Ian one of the things that you touched on was the um you know how for example content instead of requiring one FTE now the ft is spending their time qaing and editing and actually um this has been uh I think Peter touched on um you know the the worry about jobs and consideration around that right um what I'm finding though actually is that uh most folks are now turning into editors right and directors right uh as opposed to having to execute on that one really good example of this is that Jeep Wrangler the company had recently done a ad campaign uh which was fully generated by AI right uh but when you look into that it has a style it has a it has a particular features that you could not tell is being developed by a mid-journey or or something else right and that's where I think the creativity of the ftes actually come in where that's where uh some of the editing and directing skills come in where again these are just tools they need to be able to use these tools effectively uh disposal 100 love to like hear your thoughts um in your day-to-day observation the style line with what you're seeing um any other opportunity that you might have seen based on something which is discussed yeah what uh yeah uh just to compliment uh What uh Moses asked about there's a lot of good uh discussion good uh task um use cases that are happening uh in the consumer space uh what I see is um is the the customer journey and uh um in the past we couldn't track that customer Journey from beginning to end uh now we can we can we can we can track that customer Journey from beginning to end and we can tie that back to the product and this is this is a quick win and then uh regroup and readjust and deliver and I think that this is where a lot of shifts happening so one particular one that uh one of the use cases that I can talk is some of the shelves that we were the products are being you know done uh are delivered and or how the storefront looks right uh maybe you know we're doing it different from geographically from west to east or or in the in the Prairies or other part of the world uh it doesn't have to be different look uh if if something that we deliver that worked for us uh it it is at your fingertip and with with with with the fingertip you can actually put it in that hey this is how it it caught the attention so some of those marketing some of those the front end uh displays are getting moved getting shipped and and this is this is a huge win uh in terms of and how we're doing it it's it's a Content that is you know that with the help of uh some of the the use cases of the the generative AI models that we're using and I can talk that this is a huge win in in the in the front pace of the the storefront and with the customers 100 and I think it's something that I want to pick up on what we just mentioned around um tracking and maintaining customer Journeys um so I think I want to connect the thoughts back to what we've been talking about um find those insights into some kind of centralized solution within Enterprise but then allowing it for others to self-serve through maybe a gen AI interface so it's almost like contributing good data into a central Enterprise um data solution but then for people who are not very technical in organization now they have better access to Insight through the Gen AI solution that um now we're seeing right so um we have a few slides on the page but I think we've touched on various aspects of it already but essentially this we have three use cases that we selected for the audience just for inspiration and these are some of the use case that we are seeing that's gaining traction and and um and attention in the market um so the first one is really around um allowing better personalization because of now with more um original content that's being generated by marketing teams right so that can provide a much better personalized experience to your customers right so if we move to the next one um it touched on uh your use case around how can we provide better services to customers and also back office staff as well right um so there's obviously this uh need for customer getting touched to um to ask a question or get some issue resolved but as we look at the Enterprise as a whole there's also such need internally between different teams right so if we think about a centralized IP function for example and they provide services and tools to the internal stakeholder for example an online business um a lot of their time is actually spent on issue tree arching for example right so this could also be not just applied to external customers this can also be applied to internal stakeholder as well right um and mufusa I think like you have some experience in this in HR uh space I love to like get your thoughts and uh inputs on this as well yeah so what use case I can talk about is employee experience and of course the employee experience is a big piece of my journey that how we've been treated how you how easily you can retrieve some of the information how easily that uh employee portal and all those so one use case with the help of generative AI that we can we can take from higher to retire and this is this is somewhere that when you apply for a job um before you get into the the job site uh we we start using that experience and that how did you feel how how what was the experience and then you know when you get to the interview and the next at the hiring process when you become an um an employee the big pieces onboarding and how good we are providing that onboarding support I think that's a big piece of your journey as an employee that you are coming in look it's not about some of the checklists that you know in my generation or previous generation that went through that hey this is the onboarding task this is what you do it's not it's not that it's much more personalized much more uh user-centric and much for the the the the the job centering at the role-centric and that's how much its div is going and also what it's doing is it's also bringing in the the content reach content and some of these contents are pulling in from different sources some of the learning um and some of the the some of the capabilities that's bringing in and as an employee we want to make sure that you have the best experience while you're on boarding yourself you have what you need to do the best you to to become you an enabler for the job and that's that's what is a big deal and in my space I can talk that we are taking advantage of that and then how are we tracking some of the behavior how we are spending your time how you are um you're being productive what is that is uh taking most of you and how are you feeling so so again uh all of a sudden the the toolman said hey it's time for uh for a meditation and let's do three minutes of meditation and uh let's do it or let's text yoga so so this is the employee experience that you embedded we are embedding so that you have that kind of uh experience and this is all of the help of some of the predictions some of the uh the analytics that we have some of the the use cases that we have and then where do you go from there and this is we're building us we are learning you uh and we're we're building that the model and we're going forward and we want to have that their journey and and then of course as you're correct coming close to retirement we want to make sure that you are going through the same Journey so so again this is the same model uh we are what we're calling from higher to retire and absolutely and the mobile experience whatever different platform you talk um it is it is somewhere that uh we we're taking advantage we say that and some of the rich content really helping us some of the media and the content that really help us helping us and some of the experience uh built into the other capabilities uh there that we're taking advantage and that's great thanks um for um for sharing on what can be done from um employee experience standpoint um so I think the takeaway maybe perhaps for the audience is as you think about how can this apply to your organization obviously um everyone is thinking about how can this apply to external customer experience um to make your business better don't forget there's also a lot of internal opportunities that you can start with actually I would just love to touch on that as you talk about internal opportunities um so I think we touched on the fact that we are a payments company and as such uh we are in a regulated business and one of the things that we have actually been thinking about is that you know so a lot of our customer support agents our account managers our folks who do kyc KYB Etc they need to be themselves continuously audited you know we need to make sure we ask them the right questions that they know uh they're up to date they're uh they're doing their training on time right and this should happen on a continuous basis uh because we are a regulated business so one of the places where Germany bi is actually quite useful here is because is that if you can plug in generative AI as part of your regulatory requirements and as part of your processes right and your data uh based on how your support agents and others are interacting with your internal tooling you can actually automate that those prompting of the questions every now and then right uh so for example maybe you notice that a support agent uh has been doing uh fantastic work right and you want to make sure that they continue to be able to do so when it comes to compliance and onboarding you'd be able to occasionally query based on the historical data and historical interactions that they've had right using generative AI if you can automate that so that's an area that we're particularly excited about because historically this used to be be done by a compliance officer that used to be maintaining these on a regular basis but it can actually happen on a more automated scale this is awesome insights um so just want to do a time check so we have about 15 minutes uh left to put a session um now we're getting some very interesting Insight from the panelist um what we want to do is shift gear a little bit um and get into some of the key considerations I think we've touched on a few things already but maybe I can just recap and ask maybe more software to expand on that a little bit more um so I think number one um is definitely some kind of check and balances that we need to put in place right and I think there's a question um in the audience who's asking about who owns the content that was generated by the biologan AI solution right um secondly I think the second big area is around I would say people right so um now because of gen AI people are working differently now the way they use technology is different and then I think one of the key questions that we were actually discussing and debating with the client is that should I allow my team to use it the positioning was that um if my junior staff don't go through the process of creating content they won't be able to learn and also they won't be able to see what good looks like hence they cannot judge what should they select from jamf AI so that's one big consideration around people around upskilling trading or should or should not be using gen AI right so maybe we start with those two and love to hear your thoughts on those two areas so what kind of check and balances would you recommend to put in place how would you think about up schooling and training people in the context of gen AI right and then we can probably look at some of the key questions that's coming from the audience absolutely I can get started um so with regards to upscaling and uh using these such interfaces um we have been very open in our company actually I think most uh members of in our company have a chat GPD Plus subscription right uh um and that is primarily because we have we have seen the benefits of uh of just uh testing and learning on its own on their own right uh but there are certain use cases where we are um extremely careful about especially as I mentioned compliance uh being one of them um primarily because folks who use uh tooling uh many of them unless you're in the industry don't actually realize that all of this is based on historical data which can be as late as uh 2021 right um I think the latest 10gbt models were trained up to 2021 uh I believe right so uh the information you get might actually not always be the most accurate or most up-to-date uh based on the information that you have so that that was one of the key things that we've had to uh we've had to teach our our teammates as well right uh the other things are that uh I think that uh I I don't want to be too uh you know uh one too much right but essentially essentially when it comes to pii data and and uh and just consideration in data privacy right uh we do have to realize that um models are feedback loops right and if you don't if you don't own the weights you don't own the data right and you also don't own the uh the inputs of the things that are going in there uh right so if it is customer sensitive data company sensitive data right I would be a little bit more diligent right to understand how and where um this is flowing through so we are careful about those things but when it comes to General use cases um you know we are very open to adopting new tools because the rise in productivity is just uh enormous enormous right um so no doubts about that um yeah so hopefully that that gives I would love to hear from officer as well on how he thinks about it thank you more and uh I can talk probably from the legal ethical and the policy point of view uh definitely uh we we try to be responsible uh in development and some of this capability comes with um not only comes with some of the tagging some of definitely there are always we have to follow certain regulations so yeah after seven best practices and we're not uh different and what I think the aspect is you know it's a it's a new way in your tools and uh way in what works what you know the model that probably works for us works for made out of work for you and vice versa and definitely we have to weigh in the pros and cons and and that's how uh we uh we we cannot look at um I know more talked about the DVD subscriptions and this is kind of it's just kind of vibe where again with caution also it's you know who's who's getting on to it and some of some of the the bigger platforms right uh snowflakes you talked about and and post GRE so look uh there are endless opportunity it's it's all accounts to you that how you want to use it and how you want to take advantage of the feedback loop got it so um one last question before we uh start wrapping up um would you allow your team to use uh gen AI absolutely awesome absolutely um so I think the last section I know we are about 10 minutes away from uh end of the session uh love to spend some time to talk about like how to get started right um so what are some of the minimum um to requisite from a technology process and people standpoint um that from your perspective that organization started me to consider now if they want to jump in with uh maybe not both maybe with one feet so people and I we can start with you first yeah and move faster yeah yeah so I would say it's important to be very focused about which use cases you think um you know could benefit like which processes within your organization uh would be areas where you can augment your your staff um the other thing is I would say that in in many cases with regard to like the normal ways of working um we're going to be seeing a lot of new products being released right Microsoft co-pilot is coming out Google has released um a number of products for their for workspace um so I think uh we will start to transition in our normal work to a kind of an AI augmented um you know way of working uh when I look at specific processes uh let's say call center uh call center automation is a great example where there's a ready business case where you can automate some inbound interactions or another example would be on the marketing side uh generating ad copy or generating advertisements um like these are areas where you can kind of get really focused around a specific process that can be augmented and then understand the data that you have so part of getting ready is also doing a bit of Knowledge Management really really understanding what that process is breaking it down understanding what data is available and then from a technology side there's a lot of opt-ins right there's the there's many providers coming out with with models there's different types of models I think I saw a question in the Q a about like how do we make sure that those models are secure in a business process you know they're not hallucinating or generating uh or you know they're not error prone you know that's part of the assessment work that we would have to go through right like essentially with the data and the process in mind the use case in mind then there's a bit of a model evaluation and uh and also kind of configuration around the chaining of multiple models or how we evaluate those outputs to make sure that it's actually going to achieve that business case that you're looking for and then the rest is just you know software development um you know it's a bit of engineering and it doesn't in many cases because these models are off the shelf you don't actually require data scientists or machine learning Engineers but you do require developers to actually compose and integrate before we start any inputs or anything to add to this yeah so so again um without making it uh complicated so very straightforward right um I would say that uh to get started you really need to have where do you want to be and you know where you are right now and I would I would absolutely start there and based on where you want to be I think it all depends on this one of the models that you'll be you are looking at what is that without any ambiguity make it simple make it make it clean um there's there's many ways that you can you can do it uh I would I would start very simple somewhere and then start loading again and then making it making it easy user friendly and uh where this is look you know your target audience who's that your target audience that who are you catering so to me it's uh Ian and team it's it's much more simpler we started something and then definitely the capability uh what you see uh something right now it didn't start five years back so again the journey has probably it's a big uh reach and content based and so for us I think it's it's it's it's somewhere that I would invest heavily got it and what I took away from that is actually um if I heard it correctly is the discipline of product design and development need to continue all right um and we don't need to do anything different now just because um a lot of the the the the due diligence in terms of engineering processing me too we continue to do that better absolutely um before we move on to the last section love to like hear from any anything to add to this no I think uh Ian you you touched on exactly right um you know uh the the things that I tell my product teams is that we just stay the course the only difference in our thinking is that whenever we're trying to solve a user problem we try to ask ourselves um Can this benefit from a probabilistic solution right as opposed to just a deterministic software right and I think that's the only shift in thinking we've had right and that has actually unlocked several different like interesting things right uh but typically we just start extremely uh small and simple which is just by looking at the problem that we're solving and asking ourselves as we're in the solutioning phase can this benefit from a probabilistic solution got it this is great um I really appreciate your insights and uh input into this um so I know we're about three minutes away from wrapping up the session I know there's a lot of questions that we could not get through um in the chat so we'll try our best to make sure we um follow up um and and find a way to respond back but maybe perhaps um before we leave the session love to hand it off to admin just to quickly share about if you are thinking about um discovering opportunity to apply gen AI how can we help yeah so we are offering a generative AI lab construct right so your peer organization is thinking about use cases or you need assistance in looking at your business and and determining where the highest impact opportunities would be uh this is a great place to start um so we have a multi-module series the first module is focused on use case identification and prioritization um and then the the second module is really about feasibility so taking those use cases and in the room with many vendors um you know from from various big tech companies or many of our partners we can actually start to break them down and look at opportunities for an actual larger scale up so if you're interested please reach out to I believe the AI Institute to the contact information you have for this session and uh we'll get you connected in to that and this is great so before I pass it back to Aisha to help us to wrap up um Muhammad and really appreciate your time and this is invaluable to hear your thoughts and observation in the market and I hope the audience can take something away from this as well thank you very much thank you very much glad to be here and thank you to all the panelists as well as to the moderator for today's session as you'll see on screen we do have an exit poll just to get your sense of uh takeaways from this session and also um what you saw and reflected in terms of conversations moving forward this is not the last of our generative AI series next week we'll wrap up one more session on genitive AI as applied to energy resources and Industrials we're happy to have both boost power and Ontario power generation who will be participating in that conversation so please stay tuned for that as well as our upcoming webcast on our regular programming schedule which will be a data monetization followed very shortly after a two-day Gap and Then followed by web3 and computer vision as always we thank you for your attendance as well as your loyalty for the webcast series and also look forward to your feedback moving forward so with that thank you all again for staying with us throughout this conversation and feel free to reach out to us if you have any questions moving forward and wishing everybody a great rest of your day because we are at time but please take a few minutes to complete our exit poll
2023-06-05