What you'll learn
- Deconstructing GenAI myth vs. reality
- How to apply GenAI in small business relationship building, customer service and expanded go-to-market opportunities
- Leading practices for mindset, prioritization and approach
Transcription:
Kathleen Woodard (00:11):
Well, hello diehards. It is wonderful to see everybody. I think that was a very, very big introduction and I am probably not going to answer every question that you have around generative AI and banking and small business. But what I am hoping to do over the next little while is spend some time talking about how Microsoft is thinking about generative AI, how we're thinking about this in particular with a small business segment. I'm just going to take a minute to introduce myself because as Stu said, I'm Kathleen Woodard. I lead the America's Industry Advisor team for Microsoft. So we are part of the worldwide financial services team, and I have a team of X bankers, X consultants who are an investment from Microsoft, and we are targeted against the largest banks in the US, Canada, and Latin America. So one of the benefits of this kind of role is that we get to be able to see the opportunities as well as the challenges and help to think about what everybody in this room has already talked about over the last two days.
(01:13):
How do you find the right kind of technology to enable the change that you need to see? How do you think about it in terms of driving efficiencies and productivity and how do you think about it for big eye innovation? So my interest in being at this conference also goes a little bit deeper because I will tell you that while I've been a few years now at Microsoft, I'm a 26 year banker, ex-banker, recovered banker, and I started in banking almost 30 years ago last month in the small business segment. So I started in a bank, I was a credit officer, margining, receivables and inventory. And in the days when you had to do paper-based credit submissions. So for those of you diehards who also remember Lotus Notes, you will know what I'm talking about, right? And those were even in the days when I had a manager's discretionary limits of 250,000 man has the day changed.
(02:12):
So not only am I here to talk about AI in small business, but I am passionate about the possibilities of what we can do for this segment. So over the next little while, I'm going to talk about a few things. I'm just going to, oh, it's cutting. We're not loving this. Okay, randomly cutting out. Okay, let's try this. How's that? Better? Okay, awesome. I'm going to talk a little bit around the landscape of small business banking. We're not going to go deep into that. You've been talking about that and hearing about that for the last two days. But what I do want to talk about is how the introduction of AI has changed the landscape for all businesses and all people on the globe. And I'm going to do a double click in terms of a couple of use cases that we're seeing resonate across the global landscape.
(03:02):
And then finally, and almost reinforcing what you just heard in that previous session, I want to spend a couple minutes on the approach because as much as we talk about technology and the introduction of technology and getting it into production, that is just the beginning. The hard part is actually the change management, the adoption, and getting people comfortable with that. And we all talk about that and we all nod our head, but it is the most difficult part. So forces at work in terms of small business banking, I'm not going to drain the slide, but I would tell you that there's a few things that we're seeing pretty consistently. So the macroeconomic environment and geopolitical stability is obviously having an impact, but the longer lasting technology shifts and the technology integrations and innovation is having a huge impact. And so what we're seeing across all of the banks is this huge focus on cost efficiency, on optimization pressures and this balance between high tech and high touch in terms of advice.
(04:06):
And so in every session we've actually talked about what are the possibilities? How do we enable our business bankers to provide better advice and to be where the client is at for that particular moment? And one of the biggest things that we're also seeing is this proliferation of data. We're seeing increased fraud, we're seeing increased customer demand and heightened expectations around the experience. And then we're also seeing increased competition because of the FinTech ecosystem. And so with all of that, it is a huge opportunity for us to rethink business. And when we are in front of bankers across all the different geographies, we are hearing a constant refrain of, Hey Microsoft, that's great. You're doing amazing things. We're spending a lot of money with Microsoft. How can you help us think about enabling the existing investments that we already have in order to solve real business problems and to help us think bigger? So we're going to spend a little bit of time on that. You have all seen this before. Headlines are big, there's lots of hype. It's been two years now, two years in November since the launch of Chat GPT. And so how many of you when ChatGPT came out, how many of you started experimenting with it right away?
(05:26):
Most of the folks in this room, and do you remember the headlines? It was both Doomsday AI is going to destroy the world. We are all going to die and we are going to be subject to robots taking over the world. And it was these massive proclamations about what was going to happen. And if you think about it, I want to frame this conversation today with a great quote that's attributed to Bill Gates. I'm not sure if he's the actual origin of this quote, but he talks about how we as humans have a tendency to always overestimate the impact of technology in two years and always underestimate the impact over 10 years. And I think we're two years into this motion now. And I think reasonably for those of you that have experienced ChatGPT have played with generative AI, how many of you have had your life change as a result?
(06:21):
I think that the answer in 10 years, if I were here and answering that question, I'm pretty sure most hands are going to go up. And that's something for all of us to think about. How does that change the way that you work within your bank? How does that change the way that you're going to serve your small business customers? And what does this mean for the future of the economy? So a couple things. What we are seeing is that technology is always a force multiplier in terms of impact on the GDP. We know that these introductions of technology have had this massive change, but now with this introduction of generative AI, the pace of technology innovations is massive. I'm relatively new to Microsoft, but I can tell you there's folks that have been there a very long time and even the diehard technologists are finding that the introduction of new copilot capabilities of new technologies, it's becoming so fast that even the experts in this field are having trouble keeping track of it and understanding and being able to wrap their hands around it. And so if they're the experts, how does everybody else feel? The reality is this is moving at such an incredibly fast pace that it's hard to keep up with, but that does not mean that we get to ignore it or wait until things slow down a bit.
(07:48):
So I think most folks are familiar in this room with just a very brief history. Obviously artificial intelligence has been out and around for a very long time. By the way, I don't know if any of the diehards here are from JPMC. No, we didn't. They're not the diehards. One of the first usages of artificial intelligence was from JPMC. Do you know what year that was? Any guesses? 87, right? And so I think you all know machine learning AI banks have been using this technology for decades. What is different is that as we've started to go deeper in terms of the neural networks and think about deep learning and the ability of these neural networks to mimic the human brain and so to mimic with nodes and connections, the synapses and neurons of the human brain, and with the ability to actually be able to extract data, to be able to predict text and to be able to summarize and generate net new content. So rather than the old rules of traditional if then technology and some of those brittle machine learning applications that were really difficult to anticipate all of the different scenarios, generative AI is completely changing the landscape in terms of the data sets that it can reason over and to be able to generate net new content.
(09:26):
So this is now advancing into the era of copilots. And if you think about it a copilot and how Microsoft talks about it, it's just simply a powerful assistant. It helps streamline operations, it helps enhance productivity and supports decision making and all while requiring that the humans at the center of that experience. And so a couple of things I want to bust a few myths. First of all, what copilot is not, it is not a replacement for humans. We think of it as a copilot. It is never an autopilot. Copilot is really an augmentation for human capabilities. It is not perfect or infallible. Copilots will hallucinate. There will be mistakes that are made. So human oversight is still necessary, nor is it a one size fits all solution. So if you think about a copilot, it is not something that's going to be able to be one solution for all business problems.
(10:28):
It actually requires targeted specific work streams in order to enable the benefits. And it's not a static tool. So it is continuously evolving with updates, improvements to better serve the users. What it is also not is a security risk. And I think probably those of you in it and those of you, your IT partners are probably going to disagree with me on that. But the reality is that copilot handles lots of data, but it's also designed with robust security measures. So security measures for data privacy, data encryption and access controls to sensitive information. So copilot, I think of it like this. I think of it as a, you have a super smart intern, so I don't know how many of you ever had an intern working with you. Did you ever take their material and go and present it to your board or your senior executive team without looking at it?
(11:24):
It's exactly the same thing with copilot. You have a very smart intern that can turn around material very quickly, but you always have to have an oversight of it. And so it effectively acts as your AI assistant. It is a productivity booster decision support. It has multimodal capabilities. And by that I mean it is text, it is video, it is imaging, and it's all about learning and adoption as well as being a collaboration tool. So let's talk about the business case because at the end of the day, we are all in this to be able to justify what's the ROI. Actually have a view that there are banks right now that are trying to really get detailed to build the business cases for how they're incorporating generative AI. And sometimes that can be challenging because the whole premise of much of what this is doing is that it is driving productivity, it's freeing up capacity, but you still have to reinvest that freed up capacity into other revenue generating activities in order to generate the ROI. But what we are seeing is some pretty interesting stats. So as you can see there, 75% of workers are already using AI at work today, and that's a number and that's a survey from the Microsoft research lab, 75%. Now the good news is that people are seeing the usefulness of the tool. They're wanting to experiment, they're doing some of this.
(12:57):
The bad news is if they're doing it and you don't have a secure environment, what kind of proprietary data are they putting into some of these tools that you don't know about? So it is something to be thinking about. 30 minutes on average saved a day equating to 10 hours per week in some of the copilot use cases. And we're also seeing the McKinsey study, I'm sure everybody saw it for every dollar invested into AI, $3.50 return on investment. So Microsoft is customer zero when it comes to copilot. We have 220,000 employees and all of them have been enabled for just about a year and it's pretty amazing. So the Microsoft seller organization has 65,000 employees. And one of the stats I would share with you just to give you a sense of this is not aspirational, this is not pie in the sky. This is not smoke and mirrors.
(13:52):
This is real productivity gains, 65,000 sellers. And across that what they've seen is a 10% increase in sales pipeline. That's a good number, that's valuable. I think all of us would like to see 10% more pipeline. But what's truly interesting is that there's a 40% better pipeline conversion because the data is accurate, it's contextualized, it is in real time and it's better collaboration in order to bring in the right kind of partners. And with that, that's obviously translating into real savings. The other stat I would share with you from Microsoft is some of the call center stats. So Microsoft has one of the largest call centers of any organization. I think we field something like 75 million inbounds on a consistent basis. And what we are seeing in this year is a projected fiscal savings of $440 million from introducing some of these technologies. So the productivity gains are real.
(14:56):
It does not come without effort though in order to make sure that people are using it and deploying it consistently. So I want to share with you now what we're seeing globally in terms of financial services and how banks are generally thinking about it. And as you think about your own bank, I know that in these sessions everybody talked about AI and how you're starting to incorporate it. Some are further along the curve than others. We're seeing three different modes. The first mode most common is in wave one use cases within the bank. And so as it should be in a highly regulated industry like financial services, we are looking at the highest gain with the lowest potential risk. And so lots of use cases, and I'll talk about them more in a little bit in terms of being able to access information for employees.
(15:47):
Wave two is actually getting a little bit higher risk along the spectrum, but that is focused on copilot agents and things like in the contact center where we're enabling some of those external facing chatbots or agents to be able to assist in a call center environment. And wave three, which is the most difficult, but also the highest potential gain is around complete re-imagination of business processes, new product innovation. And that I would say we are not seeing that at scale yet with banks, but it will come and you can see that the most common benefits of generative AI are the four things along the bottom content generation. So whether that is marketing content for your small business customers, internal marketing summarization of documents clearly and accurately, code generation, the very first generative AI product that Microsoft launched was GitHub copilot, which is essentially a tool that can generate new code and what we're seeing there across many, many banks is a 30% increased velocity of code generation. And then finally, semantic search, being able to quickly identify with contextualized insights into the data structured and unstructured that you have in your organization.
(17:15):
So here is a bit of a survey. I will tell you this is not a statistically accurate survey, but this is an informal survey that we've done. We've done this twice now across all of our internal teams that are working with the various banks in order to collate the use cases that we're seeing and where we're seeing bigger impact. And as you can see, there's a few things. So first of all, along the bottom, knowledge search and enterprise GPT, that's part of that wave one deployment and wave one deployment, meaning this is about being able to access information, so using the bank's documents, internal policies, FAQs, so that your frontline and especially some of those newer bankers who are coming in, they want to be proficient, they want to do their lending accurately, they want to understand the structure and how you open up a small business account, but it's very hard to stay on top of that, being able to quickly reason over all of the data that you have internally.
(18:20):
And remember, all of this is secure and protected. Nothing goes outside of the four walls of the bank. The data is locked down, your data is your data. It only stays within your enterprise. So we're seeing that in many, many banks experimenting with that. Another one I talked about, the code modernization, especially as you're thinking about some of the core banking systems and core banking modernization, being able to leverage some of the code conversion has been and code documentation has been really helpful. The other two places that I would just call out, customer facing gen AI, copilot agents, probably the second most common deployment that we're seeing is in contact centers. And so for those of you that are servicing your small business customers through some kind of contact center, imagine being able to do what we're seeing in some of the banks, which is a call center agent not only has an assistant off the side that's giving them advice or giving them suggestions for the next best conversation, not just next best product, but next best conversation.
(19:30):
Imagine that new call center agent that maybe hasn't been around that long. You have, you have the ability to summarize that conversation right after that call. Use it from a compliance perspective, be able to measure customer sentiment and then finally be able to extract those themes. So be able to identify has something gone wrong with a product deployment and you're now getting customers that are calling in, what are the themes that you can actually pull out? So it's been really remarkable to see what's happening. Last thing I want to share, which I think is going to be incredible, we'll see how much this gets deployed, but even thinking about an example like Commerce Bank, so they, they're in that wave two, they're doing some customer facing deployments. What they are doing is actually using an avatar and so they're using an avatar to actually engage directly.
(20:28):
And so as you think about trying to lower that cost to serve for the small business segment and be able to provide them information quickly and accurately, they're putting a person, an avatar in front of that with the generative AI, the large language model off to the side that is pulling in all of the information from the company's website. This is just one use case, but the possibilities are really, really exciting. Okay, I'm going to share with you two ideas. One of the comments I just want to make, we have some amazing partners that are here in this room and pretty much every panel has talked about some version of you're going to make that decision about build buyer partner, right? So my suggestion is that this is not just a Microsoft pitch. As you all are thinking about the journey design that you're trying to address, you're trying to get a better customer experience, you're trying to drive efficiencies and lower costs.
(21:31):
This is a great place to start and so I'm not going to go through every step in the process. I think you heard John from town Bank talk about how he had looked at the whole process, but AI, traditional AI and generative AI can really make a difference in this process. So think about being able to quickly pull in information for the application, comparing and then summarizing if there's somebody that needs to be manually on top of it. Think about actually reviewing documents. So being able to automate the population of credit submissions and being able to quickly flag where there are outside of whatever kind of limits. And then maybe there's additional information that's needed. Maybe you have to go back to the lender, maybe you have to go back to the client being able to automate and quickly generate that email with the relevant information so it's actually being created for you.
(22:28):
And I'm going to show you something in just a minute. All of this makes a huge difference in terms of streamlining the process, drafting credit agreements, being able to obviously monitor coaching agent interactions with customers. One of the things that is real, this is not again hypothetical but real, is having the data, especially in this environment where you have many interactions happening virtually. So not necessarily all in the branch, but imagine being able to have the data from all of your teams to be able to understand what's the proportion of your advisors that are talking versus listening kind of sales one-on-one that rule. Imagine actually getting that data so that you can see it that is real, that's existing in the tools today. Imagine being able to see the data to be able to understand what are common themes that are happening in those conversations and being able to guide the best practice, being able to extract what is your best practice small business advisor saying and doing and being able to replicate that against your middle and even bottom performers. All of this is real today, and so this process is something that we're actively working on with a large Latin American bank. They have 35,000 employees, they have 500 branches, but as I said, this particular bank is using Microsoft technology. There are partners who will have pieces of this journey that are put together.
(24:11):
My ask of you or my, I guess request for your consideration, everybody in this room cares about the small business segment. There is so much we can do, and I know from personal experience that part of the difficulty is small business sits between consumer banking and can get dwarfed from just the size and scale of everything that's happening. And sometimes it sits in commercial banking and it also gets dwarfed based on the dollar size of the deals. But small business, everybody's talked about it. This is the fuel for the entire global economy and these are the people that are putting heart and soul on the line and we've all talked about the possibility of changing the experience. The technology is here, it's here today and now is not the time to wait to understand it and embrace it and experiment and think about it. The time is now.
(25:04):
Okay, I'm going to show you if this works. I'm going to show you one more possibility. And so this is again what I'm excited about. What I've learned from moving into banking. Maybe some of you already knew this long ago, but I moved into banking into a technology company and I'm telling you, a lot of the technologists are great salespeople because you listen to it and you're like, that is amazing. And then you realize, but we've never done it. It's never been built. And if you only bank will invest many millions of dollars, we can bring this to life. That's a bit of a tough thing when you walk into it from a bank where you're very much about, let's make the numbers real, I'm about to show you something that is real. Every piece of it that you're going to see is real. It is out of the box.
(25:49):
So this can be implemented in one week. This is is a SaaS solution. This is a copilot SaaS solution that I'm going to show you. This is not a long-term build. Okay? I'm going to give you a little bit of context for this. So what you're going to see here is we've got a client that is coming in Veronica, and she is sent a long string of emails going back and forth. She is the CEO for Client Quantum Connect and she is interested in $250,000 term loan. She wants to talk to her banker Jeremy about this and wants to be able to have a discussion and wants to bring in her CFO. Abby Moss. Abby is not a client of this particular bank, so there's no audio on this video. I'm going to let it run. I'm not going to stop it and it's going to be hard to see, but I'm going to give you a talking over of what exactly is happening. So let's see if this works.
(26:52):
Okay. What you're going to see up there right now is the bankers using copilot. It's pinned up at the top and in this situation you're starting to see the summarization of the multi-thread email so quickly summarizes the information. It cites parts of the email that it's pulled out so that you know that it's accurate and there's no hallucinations. It's also pulling all the recent communication from Veronica's emails, teams discussion and even servicing issues so that the banker's up to date. You can also share this information out to teams. So as soon as you have this opportunity and maybe you want to pull in the deal team, you want to be able to understand how we bring in some of the investment bankers. Maybe there's some other members that need to be pulled in that can easily be done. The other thing that you're going to see is that it is able to save information into Salesforce.
(27:47):
Many of our clients are Salesforce shops. This particular bank happens to be Salesforce because there are out of the box connectors that are built native into the product. With Salesforce integration within two clicks, what you're able to do is create an opportunity. So I don't know about you, but in some banks what we see is that Salesforce is wonderful, but there isn't a ton of adoption or it's very challenging to get the adoption. What this is doing is actually building that in so that you stay in Outlook, you're not having to transfer. The other thing you're doing is in this situation, they're pulling in data from outside systems. So we were able to see there that this was actually connected into the transaction system. It showed that KYC was not complete for this particular client, and it was able to flag that so that you complete it before you went on.
(28:41):
The other thing I want to pull out, I know we've kind of walked through lots of this, but all of this is permission and role-based. So because this is a role-based app permissions that you have in Salesforce today, interact with the data. It's exactly the same thing here. It also showed the creation of the new opportunity. And then the last piece that you're seeing here, which I love, is that as you are responding to the client, it's drafting an email. I use this technology every day. This is not a joke. I use it all the time. I think I write pretty well, but I'm telling you anytime I have to respond, do you ever get emails where you have 10 different people weighing in on the email and you're like, oh, I got to read this and analyze and figure out what it's doing. One button, it summarizes what the heck's going on, and two, push the button.
(29:34):
It's creating that email as a potential response. You always have to edit it, but it's amazing. It literally saves time. What I love about what happened here though is it's drafting that email interesting, but it also picked up the context of the conversation. It also looked at the banker's calendar to identify slots that were open and it actually responded with the email with possible meeting times. All of this is out of the box. All of this is real today. Okay, I'm almost out of time. I want to share a couple more things for those of you that are going, that's cool. My bank will never let this happen. It's too risky. I don't know if anyone's thinking this. I'm seeing a few smiles, so perhaps that has crossed your mind. I want to tell you a few things. We've already talked about the data. The data is protected.
(30:25):
It is yours. It is never used for training the models. And Microsoft is the most attacked organization on the planet in terms of fraud. And our number one priority is security. And what you will see through this is that this is actually a SaaS solution that is operating in your own protected environment. And it's also, all of this is deployed with our responsible AI standard, and we have the standard in the framework that we've been putting in place since 2017. So what you will see on there, and what we're starting to experience is that as every bank is thinking organizationally, how do I make this real? How do I put this in place and how do I put the governance in place to make sure that I'm de-risking this application? Responsible? AI has to be at the center of that. We have standards. We work with banks to provide them what we have done. We're very open about that in order to enable that for each of our clients.
(31:30):
Last thing, the AI implementation. So five factors. You will not be surprised. We've talked about it today. Number one thing is organization and culture. So what's the leadership design principles? What's the operating models? What are the skills? What are the resources you need to put in place? Business strategy, AI should be in support of the business strategy. AI is not a strategy, right? It is about what customer segment are you targeting in order to grow your business, and how does AI enable that applied AI strategy and experience, expertise, repeatable processes to create sustainable value, your tech and data strategy. So what's the data and the infrastructure to fuel and run AI operations at scale? It's interesting and valuable to do the use cases. It's completely different when you're talking about servicing hundreds of millions of transactions and tens of millions of clients. And then finally, AI governance, the processes, the controls, and the accountability to govern privacy, security and responsible use of AI. Here's my last slide. This is what we're seeing in the market right now. We see most banks exploring. We see few banks experimenting. We see even fewer banks scaling this out.
(33:01):
But what is really the opportunity is that last piece innovation. And if I had stood here a year ago, I would've told you we're not really seeing much innovation maybe in the partners, but not in the banks. It is happening today. I'm seeing it in Europe, I'm seeing it in Latin America, and I know the US is a very different market. You've got different regulatory constraints. It is big. But I'm telling you, there are banks that are completely redesigning their entire operating model, leveraging AI and generative AI to change how they deliver products, how they serve customers, and how they think about having an assistant, a chat assistant, to be able to completely change the customer experience. And so my ask my hope for all of you today, I know that everybody's in a different stage. I'm not a technologist. I'm a business person. And some of this is new and some of this is at times it feels overwhelming.
(34:05):
But each one of you, I hope, is working with AI. Are you experimenting with it yourself? Are you thinking about how this might apply in your own business? And how do you help be a voice for your bank to truly be that bank of the future that's going to survive? Because there are going to be winners in this model, and I truly believe that the winners are going to be those that as a part of their strategy, have really figured out how to unlock capacity and productivity because of generative AI. So thank you so much. I know I've gone a little bit over, so I don't think we have time for questions, but I will talk to you afterwards if anybody wants to talk. I'm happy to do so. Thank you very much.