So today in this session, I'm going to talk about, you know, how, so we have already seen that, especially during the past one year, there has been a big rise in generative ai, especially with the coming in of chat, GPT and other lms. So in today's session, we are going to talk about, you know, some of the important use cases of generative AI and how it can help small, medium sized businesses to sort of accelerate their development processes.
So in the agenda for today, we are going to talk about, you know, the rise of generative ai and take a look a little bit at the history of generative AI and what it is. And then we are going to talk about four or five main use cases of generative AI and four or five different business sectors, which generative AI is going to be very helpful in.
So we are going to talk about, you know, customer operations teams and sales and marketing teams, software engineering teams, and so on.
And then towards the end we are going to see what are some of the risks associated with gen gen ai and what are some of the good things about it. So now first starting about like, you know, what generative AI is actually, so it's not, you know, a very, so if you talk about AI in general, so it's not very super new thing.
In fact, it has been there since, you know, many, many decades. And you know, in the beginning we had started from really if you learn about machine learning. So in the beginning it starts from really simple ML models and gradually becomes more and more complex and more and more detailed. So on one end of the spectrum, you had, you have very simple ML models based on, you know, linear regression or logistic regression and so on.
Then there are more complex ones, which are based on neural networks. So neural networks, if you talk about it in very simple works.
So neural networks are fashioned around how human brains work. So like in human brains, we have many different layers within the structure of the brain, and those layers constantly communicate back and forth between each other. And that is, you know, a very high level and simple model of how neural network works. And generative AI is one step further ahead and it belongs to the category of what we call less transformer models. And I will not go much detail into that because that's a topic for a separate day.
But yeah, so generative AI is more on the very, very advanced spectrum of AI and based on transformer models. And some of the breakthrough tools which have come about in the past one or two years are of course, you know, TRA GPT and also stable diffusion.
And then many other tools which have arisen because of this advancement in l lms like perplexity and you know, metas, LLM Lama and so on. And they kind of like help a lot in automating routine tasks and also in creative and diverse.
So if you may have already observed many of the LinkedIn messages you might receive these days, so many of the emails you receive nowadays, you can look at them and kind of like see that they are, you know, created using creative ai. So on one hand it does help you to automate, you know, routine tasks. But then on the other hand, of course, you know, a lot needs to be done more for it to match, you know, the, you can say more cre, more creativity level of real humans. So what about the economic landscape and some of the future trends?
So in the past couple of months, there has been a lot of investment going on in generative ai.
And just in the first couple of months of 2023, there was $12 billion of investment in gen ai. And ever since it has, you know, become a very widespread and widely used tool and a lot of investment is concentrated in the Silicon Valley and other, you know, mega IT hubs all across Europe and UK and USA.
Okay, so now talking about some of the interesting use cases of generative ai. So for example, how the topic of our conversation was how generative AI will change or reshape businesses. So if you talk about a small startup or a medium sized enterprise, it has different divisions in it. One department is sales and marketing department, another one is software engineering and testing. Another one might be, you know, customer support or customer operations like we are going to say, so how can each of these departments sort of benefit from generative ai?
So we are going to talk a little bit about that. And we are also going to talk about, you know, what is going to be the long-term impact of, you know, coming up of generative ai. So a first and very important, and you can say a very lucrative use case for generative AI is in improving the customer operations. So what do we mean by that? So if you take a look at this image right here on the right hand side, so you see a person here who is in the, you know, with their shopping bags and they are about to check out in the kiosk.
So generative AI can hugely help in automating these kind of self service interactions because generally what might happen is that if you are, you know, in retail stores or shopping for groceries or those kind of things, you will observe that when you reach the, you know, cash counter where you're supposed to get your bills at that time, you know, you, you hear a bit, you know, a repetitive kind of conversations that hey, if you buy things you know of 20 euros more, you are going to give you, you know, 40% off on your next purchase.
So those are very standard things which, you know, the person at the cash counter or the cash kiosk would have to say to you. So these kind of, you know, repetitive interactions can be very easily, you know, automated using generative AI because you know, the customer can interact with a human like chatbot and can also respond to more, you know, complex queries. So for example, if you are a routine purchaser from, you know, a clothing brand and if you want to inquire more things to the person regarding when the next sale is coming up, et cetera, et cetera.
So all these conversations have very fixed responses and therefore they're very good candidates to sort of be automated using Trinity ai. Okay, so now in here in the next image you see customer agent interactions. So you see two people here and one of them is the customer and another one is the agent who is sort of, you know, a customer support agent, customer operations person.
So the human agent can sort of use AI developed call scripts and real time assistance.
And you know, this has been seen in the past couple of months that there are AI tools which have been invented and which have AI startups which have got a lot of funding which provide, you know, AI enabled sales calls. So if I am a salesperson or I'm helping out a customer, you know, with any of their troubleshooting or something, then you know, AI can sort of help me to sort of craft a very good script to sort of interact better with the customer and also receive real time assistance and suggestions for responses during, you know, whenever I'm interacting with the customers.
So if I'm a customer support person, I'm talking to you, I have my AI helper, ai co-pilot on the side and while I'm talking to you, and then you give me a response.
So during the time you are giving me response, my AI chat bot is continuously processing what you are saying and you know, what could be the next best thing which I could, you know, suggest to you or recommend to you. So this is how, you know, customer agent interactions can be sort of enhanced to a lot of degree.
And another very helpful thing is that if I'm a customer support person in general, I would be dealing like, you know, 10, 20, 30 customers or more in every, each and every day and each and every day to keep track of, you know, all the past history of the customer is very, you know, a bit of a drag for me and a bit of a tedious tasks you can say. So for that journey way I can help me by telling that, hey, this is the whole purchase history of this customer. These were the times, you know, they applied for some refund or these were the occasions they gave a bad rating to your product.
And then it can help me tackle my customer interaction much better without me having to go and find out all that information by myself.
And then of course it will help in agent self-improvement as well. So if I'm a customer support operations person, then I can obviously, you know, go back and do my homework and receive a summary of the whole conversation I had with the customer. And you know, I, it can give.
So you might have seen that there are many AI tools which, you know, record the whole meeting transcript and will give you that, hey, this was the summary of the whole meeting, these were the problems which were surfaced by the customer. These are the next steps, you know, from this meeting. So it can help a lot in agent self-improvement as well. So agents can sort of use, you know, automated and personalized insights, which are, you know, AI generated to sort of, you know, lead the conversation better in the future.
So if I am supposed a product manager and I am talking with, you know, customers, and so AI copilot sits next to me and constantly records our conversation and then at the end of the meeting tells me that, hey, okay, these were the three problems customers mentioned and these are the two next steps which I need to do, and you know, these are the, this is how I should take the conversation forward during the week.
So it helps a lot in agent self-improvement.
Okay, so now in what ways does it help the sales and marketing teams? So first is that it helps in creating, you know, sales and marketing strategy. So just creating strategies of how our marketing roadmap looks like for the next six months or three months or 12 months. So with that, generative AI helps a lot because generative ai, if you run a generative AI software on a large amount of, you know, data available over the internet, just like how perplexity AI does, so it can help you gather, you know, market trends and customer information from a lot of unstructured data.
And why is that a very important problem? So that actually is a very important problem because, you know, there are two kinds of data. One is the structured data and another one is the unstructured data.
Structured data is the one which is, you know, very, very systematic with properly defined fields in the database where everything is stored, you know, the data types and formats of everything. And it's very, you can say, very efficient to manage that and perform operations on that data.
But the surprising part is, is that nearly 80% of the data in a given company is mostly, a lot of it is unstructured data. By unstructured data, we mean is that data doesn't really have a fixed format or you know, any sort of predictability. So for example, if you are, if you are a company who has presence in many social media channels, for example, you are booking.com or airbnb.com. So then you have an Instagram page, Twitter page, and then there are blogs and then there are, you know, thousands of marketing campaigns going on with you are like, you know, many, many different data types.
Some are images, some are videos, some are you know, blogs, some are tweets, some are this and that. So it's like, you know, a huge variety of data in that. So for managing all the unstructured data, generative AI can help you to sort of make sense of a lot of that unstructured data and gather information related to market trends. And here the data sources, you know, as I mentioned, may include, you know, news and customer feedback and this and that, and Ative BI can help you to analyze all of it and give you a very big picture in a very quick summarized version.
And then based on that you can craft better marketing and sales communications. And then it also helps with awareness. So customers see campaigns tailored to their segment language and demographic. So for example, you might have seen that Google, if you go to, you might have seen those Google doodles, like sometimes they'll be like, you know, some events happening in the world, maybe the Olympics are happening or maybe you know, it's a world Science day or something like that.
And then sometimes if you go in a particular geography, for example, if you go in certain countries, like if you go to India for example, or if you go to Germany or France, sometimes you will get Google doodles which are very specific to that country or region. So similarly, these kind of very region and demographic specific kind of marketing can be done with the help of ative AI very, very quickly. And at very big scale, each and every campaign can be tailored according to which region do you belong to.
And then on the customer's end, it helps you to sort of gain a very vast data related to, you know, able to compare products and you know, have dynamic recommendations and so on. So like for example, now if I have to, I'm traveling somewhere if I have to book a hotel anywhere. So during way I helps me to sort of compare, you know, different products or you know, which, which airlines I should be using to sort of make this travel.
What are, what is like, you know, the safety rating of this airline, how many times does it land on time? So all this kind of was data, which is right now spread everywhere on the internet. All that can sort of come together in one place and help the customer to sort of make a better decision of which product you know, they should be going ahead with.
And then of course it'll help with retention. So in the image you see a person, a human is talking to a robot or you know, a chat bot like thing.
So you might have already seen that in many number of places in food ordering apps or in Amazon and everywhere that you know, it's a chat bot, a generative AI infused chat bot, which is actually doing conversations with you and you know, providing you refunds or not providing you refunds and so on. So it helped with customer retention as well. And now coming to how AI would assist in software engineering.
So if you may have experimented using Chad GPT for writing code in the beginning you might be very amazed that wow, you know, journey AI can write really good code and this and that, but then eventually down the line you figure out that that code is not directly usable and sometimes it'll hallucinate and give you a bit of like, you know, incorrect code, which you can't directly use, you know, in your code base.
So all that is there, but despite that, you know, directive AI can help a lot in inception and planning.
So especially if you are an engineering manager or if you are a product manager who is managing a software product, then it helps you take, you know, 200, 300 more than that items which are there in your project backlog. And then it can help you analyze all those items from your product backlog and help you to prioritize that based on, you know, user feedback or how desperately users want a certain feature and then be able to plan it in your next, you know, software sprint cycle. So certainly it can help with planning and then it can also help you with system design.
So the engineers can sort of use derive AI to, you know, create multiple architecture designs very quickly. So like for example, if you are designing a video streaming platform or if you are designing a newsfeed platform and so on. So journey of AI can help you come up with a system design and sort of very quickly iterate on that at least. So journey of ai, if it can't write, you know, the end, you know, fully finished code, it can help you do a lot of, you know, preparation around the whole software development process itself.
And then of course it can help a little bit with the coding, so it can help you make like, you know, very basic drafts of how your code might look like or it can give you a bit of strategies around how to, you know,
Excuse me that I need to interrupt the, the, the session, but we are already on time for the next one. Okay, cool.
If, if you can wrap up the, the, the concept of your presentation, that would be great. Thank you.
Sure, sure. Just wrapping up in two minutes, so, okay. So it can also help you in writing, you know, testing scripts for your softwares. And then besides that, it can help with maintenance.
So, you know, writing code is one thing, but maintaining it on cloud is another big thing. So it can help user sort of analyze system logs and all those things. So I'll just skip like a one or two sections. And of course these use cases you might have already seen writing, you know, fast marketing copies and you know, generating content and it, it also helps in retail. So just like an ending thing I would mention, what are some of the risks and rewards?
So some of the risks associated is that of course, you know, a genetic AI is trained on a lot of, you know, proprietary data of people which they shouldn't be training on ideally.
So that is still a big problem. And you know, basically the intellectual property problem and the privacy problems is very, very big. And besides that, sometimes, you know, there might be biases introduced in the model.
So for those things, you know, those are some of the risky things which are an outcome and part of generative ai, but then a way, and of course people getting, you know, unemployed because of generative ai, but the governments and also the corporates should come together to retrain people and re-enable them to use regenerative AI and, you know, perform better. So yeah, timely retraining of the workforce should be done as well.
So, so I think that's pretty much it from my end. And thank you so much for joining in today.
Thank you. And please join me in thanking Ashira for a wonderful presentation. Thank you.
