This Week in Applied AI with ElectrifAi

This Week in AI, applied AI specifically. My name is Noelle Russell. I am going to give a few minutes for our friends to join. I know it's in the middle or closer maybe to the end of some of our workdays, in the middle for others, but really excited to be able to— excited to chat with all of you. The chat is open. We are broadcasting on Electrify's LinkedIn page, but we are also broadcasting on a bunch of other channels. So if you are in a position, you can type in, hey, hello in the chat, or if you have any questions as I present today, you can certainly do that. As I mentioned, I'm Noelle Russell. I am the VP of Solutions Architecture at Electrify, and it is my pleasure to share with you a couple of really exciting things. One is is that we are going to start delivering an awesome, fun 30-minute session every single week called This Week in AI. And as part of that, we'll talk about applied AI models, what's happening in the industry, some cool and fun things just in case you've missed it and get you all caught up on a weekly basis. The other things that I will share with you today is we have a really cool announcement. I am so excited For those of you who are friends of Electrify, you may already know it, but for those of you who aren't, you'll have to wait till the end, but it's super cool. So I'm really excited to share it with you. What I'm gonna do now is I'm actually going to, I'm going to go ahead and share my screen with you and walk you through a little bit about who Electrify is, because you might not be familiar with it. You might not have heard about us, or maybe my recent announcement was the first time you've heard about it. But I definitely wanted to give you an update as to where we are as a company and the things we are working on, the super fun and exciting things that we are working on. So, so great to see all of you. Thanks for the chat. Hello, my friends. So thank you so much for connecting with us. All right. So I'm going to share my screen right now. You will see a little bit of the, the live studio as I do this, but I will flip over to PowerPoint and you can see my presentation. So there we are. Hello, friends. All right. So let's take a look. I'm actually going to go right into presentation mode so you can see a little bit about this. Hopefully you'll be able to see, ask questions. It's really exciting. There are so many amazing things you can do. To give you a little bit of a history about why I got excited about Electrify, and we actually did a recent blog post about this on the Electrify web page. You can go and take a look. But one of the reasons that this is exciting is because I have actually worked in AI for a few years. I was an early member of the Amazon Alexa team, and I saw what it took to build machine learning models from scratch, specifically some of the hardest ones, natural language models, and then start trying to augment other models with that— computer vision, text analytics, sentiment analysis. And it was really very difficult and cost a lot of money, lots of people. I mean, it was a hard, initiative. It worked out because lots of people know who Alexa is or what it is now. But at the beginning, it wasn't fully, you know, guaranteed. And so I got very excited about what can we do to democratize all of that work and make it easier and more accessible for people. And so I went to Microsoft after that and worked on a team involved in Microsoft AI and Cognitive Services, and it was all about how we could do that, how we could take models, how we could leverage the work being done by machine learning teams and make it available to companies who maybe don't want to or can't invest in machine learning expertise in-house. And it makes sense, right? Because it doesn't necessarily make sense for someone whose core competency is retail to invest in a large data science and machine learning engineering team and a research and development team who's just gonna try and figure out if a model they've built is actually going to work for the company. There's companies like Electrify.ai that do this today. Machine learning as a service. It's a relatively new concept, but it's really the only way AI will really become, like the promise of AI will become real for companies in the long term. So let's take a look a little bit at what this is. I mentioned to you, I'm so excited to announce this week, we are going to start every Thursday doing This Week in AI and talking about about the models that we're working on, the conversations we're having with customers, also anything that might be in the news that's cool and fun and exciting about Applied AI and how that might be relevant to you and just keep you up to date. So it's really exciting time. This being our first one, I thought I'd give you an introduction to who we are as an organization and just like, let you see it and hopefully you'll find it. One of the things that Applied AI offers, right, is the ability to take advantage of machine learning and artificial intelligence, right? The ability to accelerate your capabilities in AI, specifically in machine learning without actually having to build it from scratch, right? That's the key word is applied. I get to apply the work of machine learning, but I don't actually have to build it from scratch or build the expertise in-house in order to do that. So what we've found over the, you know, and through all the clients that we've worked with, What we ended up finding was when we go into an account, you don't necessarily want to start with like the biggest, hairiest, hardest problem, right? We want to start with something small, something small but impactful. I always call it like the minimum delightful product or the minimum remarkable product, right? What's the minimum thing we could do with AI that would show you huge results so that you could start to believe that this is real? Because let's Let's face it, a lot of us know that there's a lot of hype in AI, right? There's a lot of people saying, hey, we can solve all your problems with, you know, an artificial intelligence solution or with a machine learning platform. And what we wanna do is really kind of, you know, get rid of all of the hype and just show a project that works with deliverables that matter. And so that's what this journey is all about, having a customer journey that takes you along and starts off with a problem that everyone has. Maybe it's in the area of sales, spend analytics, right? Maybe it's just figuring out, especially in the time of a pandemic, how much money are you spending? Where are you spending it? How many vendors do you have? How are they categorized? Do you really understand how those relationships are being used and nurtured and leveraged in order to save money for your company? This is— everyone has this challenge today, especially as our economy has shifted during this time. So what a great way to start something small, take a look at some of your procurement data and generate a result that saves the company money. It's a great entry point for AI. Another though is understanding more about your customers. Again, now we're engaging with customers more and more in a digital way. How do we learn more about them, more about what they are, more about the systems, right? Many times our companies have lots of different systems that know a little bit, a piece of who a customer is. We can use machine learning to aggregate that information together to understand the lifetime value of a customer, to understand, you know, the segmentation of our customers so that we can engage with them better. And then the second thing, or maybe it's the third, but the last use case I wanna talk to you about is around customer engagement, right? Now that we know a little bit about our customers and we start moving down this path, What does it look like to actually start knowing enough about our customers to do things, you know, like churn analysis, right? Who doesn't wanna know if their customer is leaving and how to keep them, right? How to engage with them in a new way. And AI is specifically good, machine learning is specifically good at solving these types of problems. So it's important to understand that it is a journey and it starts with a pretty reasonable small little step that generates real results for a business, your business. And then you're like, all right, let's do this. Let's see what else we can do. We don't want to try and boil the ocean. We want to get real results really quickly. And that's what our Model Factory really is all about. There are a bunch. I just thought it would be cool for some of you who don't know who Electrify is. These are some of the things that we have done. This is not an exhaustive list. There are There was actually like several pages that I was like, maybe I shouldn't go through them all in our first introduction here. But that's what we're gonna do in This Week in AI. We're gonna dive into these use cases and just talk a little bit about one or two of them every single week. And it's not specifically about us as Electrify, it's more about this is what machine learning is good at and what your company, if you're not already doing it, what you wanna be thinking about. So you can look at some of these, like just the ability to Pull, you know, extract content from contracts to do better analysis, do supplier classification, be able to figure out and understand more about your customer through profile data. I mentioned some of these use cases already, but these machine learning models, the benefit of our organization at Electrify is that these models are not new. We're not building net new machine learning, right? We've built these models, we've trained them, we've used them with other customers, and now we're leveraging that IP in order to make it available to you as a service. And so that service model ends up really turning into a deployment or a container that you can access without having to know all the nitty-gritty details about how that model is trained and optimized and tuned. We take care of that, right? That's part of the as-a-service portion of the terminology. So one thing I wanted to leave with you before we get into our cool new announcement is, how you might think about one, you know, an initial project. There's an easy way to get started. It's how I got started with Electrify. When I first heard what they did, I was like, "Oh, this is cool. Let me go check it out." And you can actually go on GitHub and see our models. You can look at our Jupyter Notebooks. You can take a look at our common data model for each one of the models and what it needs to work. But for many of you, you may just wanna be like, how can we make, How can we see the truth about how this works for our company? And what it often includes is me or someone on my team. I, again, lead solutions architecture. So one of our solutions architects sits with you and says, what's your biggest business problem? What are some of your business problems? And then we map that business problem to a model. And we have a set of models that we call kind of our model starter kit. And we start with that and, oops, sorry. We start with that and that gives us the ability to go in and deliver value right away against a need the business has today. So the way that we typically do that is obviously my team doesn't know that I'm live right now. You all know I have so much fun with these. We start with figuring out what that business problem is. We then define the data we need to solve that problem with the model. We map it to models. We map it to the model, that's the system integration piece. We do the model training and then we go through the process of generating results. And it's the outcome of this POC that makes it so interesting, right? Is our ability to generate a report. One report is your data quality report. How good is the data? How much work did we have to do to get it into a model? That's super useful. Even if you never do anything else with the POC, It's really useful to be able to know how good is your data in order to solve the business problem you're going after. This last part or the last report that you get as part of this POC is of course your model performance report. And this is where the gold is sometimes and almost always where the customers that we work, they get this report and they're like, oh my gosh, what else can we do, right? What is the next thing that we could do with you? So it gives us an opportunity to really in a very small scope walk through the process of using some of our IP, some of our existing models, train them on your custom data and have that generate a result that's unique to your business that moves the needle. And this is in my mind what makes AI a reality for companies, right? The harder part is often in, oh, there we go. The harder part is often in the details, right? Many, many times we say the devil's in the details. So I'm just gonna cover very quickly a couple of those details for you because oftentimes we have, you know, our clients ask us, okay, great, that sounds awesome, but what does that really look like? How does that really work? And so we have a couple different ways that we work with customers. One is this model inventory. So we have a whole collection of models that we have created, tuned, we've done feature engineering work with to get them ready. Then we— now we want to make— we want to align those models with business problems our customers have. Now oftentimes our customers have SIs or systems integrators or partners that they are already working with that solve other problems for them, right? Like digital transformation, cloud adoption, those types of things. And really we want to just partner with those people. We're not in the business of coming in and necessarily doing all, you know, being your data scientist team on site. If you have a partner that's already working with you in this space, we can augment them. If you are a partner, we can support you and give you a collection of models that you can add to your inventory of solutions that you can offer your customers. It's really exciting. So what you see here is the magic is down at the bottom where we build a model based on, again, that alignment of business problem to use case or to model that solves that problem. And then the magic happens in mapping the customer's data to that model. And once we do that, we're able to train the model, get some early model scoring done. And again, that model evaluation and data quality report that's returned to you to say, is, you know, is this doing what we want? Is it achieving the goal we want? And then oftentimes, once we do achieve that goal, it's a no-brainer. Usually we're asked like, okay, great, we want to keep doing this, right? So it often starts with a POC and then turns into a monthly subscription for our models, at least at the beginning. Then we start getting into some fun and interesting use cases in your specific area of expertise and domain, which might include computer vision and text analytics and call center management and all the different spaces that we've got expertise in. So it's really fun and exciting. I wanted to share with you because as I've been in this organization for a while now, looking at the way that we've decided to tackle this machine learning as a service business, I thought it was really cool and I was actually surprised that not more people knew about it. So I thought our first This Week in AI would just be a little bit about how Electrify does what it does. And of course, if you're interested, you can reach out to me, connect with me on LinkedIn, say hello, and we could talk about it. But I just wanted to give you an idea of how, how this could actually become valuable for your business. Because many times, like I said, I'm used to hearing the promise of AI. But now I really want to see it become a reality for companies. And this has made it a reality for the clients that I'm working with. And I'm super excited to share it with you all. Now, the next thing I want to do is just quickly go over, and I'm not going to really spend much time here, but this is the list of, of, you know, use cases that I've ended up leaning towards as those first POC type use cases. And they're all things that most organizations across industries care about, right? Reducing costs through spend analytics, knowing who your customer is across all different data sources, understanding how to engage with your customers, cross-selling, upselling, churn analysis, right? Demand forecasting, dynamic pricing. These are things that are in our direct wheelhouse that we have been delivering for customers for years, right? This company is 15 years old, has been doing machine learning for a long time. So build on the shoulders of giants, take the models that we know already work and make it part of your solution. It's just such an easy no-brainer for most of us who are struggling with, my gosh, how do I get into AI without buying in, you know, buying a team and creating a team and managing a team? This gives you the ability to do that. So that actually leads us right to our announcement, our ability to to accelerate more and more companies through partnerships. And we've got a brand new partnership we wanted to announce. It's so exciting. There is a new competency at AWS called Applied AI, and it allows us to really, as Electrify, to deliver these models to you in an applied AI fashion, which means you get the models and you apply them. In other words, they become available to you through an API, right? Completely consumable through, in our case, a Docker container or through AWS SageMaker, right? Just going to SageMaker. And what I wanted to do with you really quickly was take a look. I'm actually gonna escape out of this and go back to my browser. Here we go. Hello. And I'm going to just take you into SageMaker really quick so you can kind of see what this looks like. And this is what I did. When I first started understanding what Electrify offered, I just dove in and was like, "Wow, there's 65 models available through SageMaker that you can consume." Now, what I'd suggest is we get to be friends, right? We sit down, we talk about it, we talk about how these models have been used, what our successful case studies are, right? We get on a call and just have a conversation about it. But if you, you know, wanna just take a look at what's available and what's out there, you can see all of these models here. And let's just take a look at one. Cross-sell for the insurance industry. Right? This gives you the ability to read a little bit about it, look at its usage, and, and see, you know, the categories that it's in. But one of the things I really like about it is we also have the ability for you to dive into some of the more technical details in our GitHub repository. So I'll show that to you really quick as well. I should have it. Up here. So we're just at Electrify on GitHub. So for those of you who are a little bit more technical and wanna dive into some of these models and see the Jupyter Notebook and maybe test some things out, this is for you. But it's kinda cool, right? And I'm so excited to announce our relationship with AWS. I was a Certified Solutions Architect. I got their Cloud Practitioner certification. I worked at AWS as a Principal Architect for Alexa. It was, I am so excited about this. And I think that there's a lot of things that we can do together. And so I wanted to share with you, for those of you, I'm actually gonna go back and stop sharing my screen. Hey, Frank, what's up? So great to see you. And I'm really excited to have an opportunity to just kind of hang out. One thing I wanna encourage you to do is Follow us on social. So we will be doing this every week, so you'll get a chance to get to know myself and our team, but we will also be publishing content not only on our website through our blog, but also on Twitter. So I can go ahead and I'll let you know what that is now. So there we are. You can follow us @Electrify, very easy to do. We also will be doing Ask the Architect sessions sometimes when, during our week, This Week in AI, where if you have a question about applied AI, about how to use it, about where it's going, or challenges or concerns, these are things you can bring to us and we'll be, you know, we can start a panel discussion and answer it with some of the best industry-specific and domain expert architects there is. And I'm super proud of the team that we've assembled here. We've also recently, you know, happy belated International Women's Day to all of my friends out there. But we have such an incredible diverse team here. Here at Electrify. I'm really proud to be part of it. So the last thing I will share with you is really just connect. So you can connect with us on our LinkedIn page. You can also connect with me personally. As most of you know who are my friends, I am always very excited to make new acquaintances and start a conversation. You never know where it's gonna go. And if you're not leveraging AI in your organization and want to, Machine learning as a service, applied AI, what I just went through here is an easy entry point, right? We can start with something small. You can see how AI will work for you and it'll be a pretty incredible journey from there on out. So I'm excited to share it with you. I look forward to meeting you guys here every single week. We'll spend half an hour together talking about the latest and greatest in AI. It's been really fun. I hope you enjoyed it. Definitely jot in questions in the chat. If you're catching this on the replay, thank you so much, uh, for coming back and watching it. If you're catching this live, thank you for being here. This is amazing. It's our first one, my first one. Um, so This Week in AI, that's a wrap. Thank you so much for joining us, and I'll see you guys next week. And, uh, congratulations Electrify, so awesome that, uh, we got that partnership. So really cool. Thank you, talk to you soon. Okay, bye!