How Salesforce is Using AI to Transform Customer Relationships

Automatic transcript of the recording (English); speaker labels are automatic. Timestamps open YouTube at that moment.

0:00Sacha: All right, so today I'm delighted to welcome Jag, a seasoned solutions engineer at Salesforce, a company that is transforming the way businesses operate with their CRM solutions. Jag has been with Salesforce for over three years and a half, currently serving as a senior solutions engineer in the enterprise CS division. His accomplishments at Salesforce are numerous, including being part of the FY23 Peak Performance Club and presenting at Dreamforce 2022 in San Francisco. Before joining Salesforce, Jag was a principal consultant at Knox for over 13 years where he designed, managed and built complex branding, staging and audio and visual solutions for the enterprise and SMB clients. He also served as a general ledger officer at TD for two years. And with this diverse experience, Jag is truly a force to be reckoned with in the tech industry. Jag, it's an honor to have you with us today. Welcome.

0:52Guest: Yeah, thanks. Thanks for the invite, Sasha. Do you prefer Sasha or do you prefer sash the way it's on your hat?

0:56Sacha: I love sash. Sash

0:58Guest: it is. So a little bit of background. Sasha and I met in Montreal, a month and a half ago, maybe a month and a half ago, over dinner with some mutual friends. And we really had a good conversation. Then he told me about this podcast. So I'm honored to be on it. Thank you for inviting me.

1:13Sacha: Thanks for coming. So Jack, can you tell us a little bit more about your background and your role at Salesforce to begin with?

1:19Guest: Yeah, sure. So I serve as a solution engineer at Salesforce. I started off in the SMB space up until last year when I moved over to the enterprise. In our enterprise space, we are verticalized. And so I serve our manufacturing, automotive and energy customers from the core side of things and think of that core as like core sales service marketing use cases. And as you may or may not know, Salesforce has been diving deep into the world of data and AI, for which I have what we call a black belt. It's an internal accreditation given to folks that are particularly specializing in that. And for that, I operate across the verticals. So that's been my life for the last three and a half years. I take a lot of pride in talking about my nontraditional background as well. Prior to Salesforce, I operated my own company, which was called Knox Incorporated. And that was 13 years of doing a lot of different things. So that brought me to eventually joining Salesforce four years ago. And I feel like I still learn a lot from my past and I apply a lot of things from my past in my current role.

2:19Sacha: That's awesome. So you mentioned that you have a black belt, but I'm curious to know, how did you become interested in data and AI in the first place? What is your journey been like in this field?

2:29Guest: Yeah. So prior to joining Salesforce, I was actually more interested in data and analytics in general. And I think, to be honest, for every person you've probably interviewed on here, they probably said the same thing. And the reality is that before GenitWi became mainstream, there was a hot minute where data analytics was mainstream also. And we're talking about just predictive, good old predictive statistics, right? So before joining Salesforce to transition me from my previous life to this one, I went back to school for data management, data analytics, where I learned, you know, hands-on keys like SQL, R, Python. And my desire for that, honestly, was to kind of try to understand how to quantify human behavior. And the application of that is that in the CRM, but quantifying human behavior can mean like, if you are a musician, how do you read the crowd? That's quantifying human behavior, which is using your natural senses rather than using an algorithm. So essentially, my goal was to kind of learn how to build an API into someone's mind. And that took the shape of data analytics.

3:27Sacha: That's great. I also have kind of the same background, so I can really, really resonate with you here. And I'm curious to know what excites you right now. So you've talked about a little bit about Gen.AI. What excites you the most about disk current states and the future of AI?

3:42Guest: Yeah, I think you're going to hear a recurring theme in the next few minutes about like my particular views on AI. I think I'm coming to a point where I'd be curious to understand how this conversation will change in the next two years. And what I mean by that is I'm more excited about when the dust settles what remains of the mainstream applications of AI. So I know you're big on music. I see that in the background there. And I see, you know, with the conversations we've had, if we go back to circa 2000 when Napster came out, right? And everybody was like, this is the next big thing. We're just not going to pay for music anymore, right? And what happened out of that? We had unlimited, abundant access to music. But it also made us less excited and less happy. Every time we heard a new song, we valued each song less. To the point where in the mid-2000s, there was actually a rise in vinyl sales, which is antithetical to Napster. Why? Because at the end of the day, anything that gets too popular too quick hits diminishing returns.

4:44And we go back to something that's more natural to us. So I'm more excited about once this hype cycle dies, what remains of it and how we kind of apply that in a way that's more sustainable. So that's what I'm more excited about.

4:58Sacha: Yeah, that's a very, very interesting take. And I think that's going to be also a very, I'm very curious to see what happens in the next five years as well. That's a great take. Now, if we go a little bit more into Salesforce, how is Salesforce for people who don't know, leveraging AI to enhance this CRM solution that you guys have?

5:24Guest: Yeah, sure. I think it's important for us to kind of start to talk about what is Salesforce in the first place for people that may not have heard of it. So Salesforce is a B2B solution, essentially a customer relationship management solution. So if you're a business that wants to manage and quantify its customers for sales, service, marketing, use cases, Salesforce is a tool that helps you do that. It's a cloud-based tool. And from a technical perspective, it's a database with a UI layer that's user-friendly for customers. They're not building out SQL views or tables. They're just accessing a UI where they're entering information and getting some out of it. That's Salesforce. As you can imagine, a lot of our customers' expectations have changed, particularly from the service and sales side. So if you call a Bose service center because your headphones not working, you expect them to know who you are. You expect them to understand what your issues may be and resolve them as quickly as possible. So we're using AI to augment a service or a sales user experience to make it simpler to do things and to amplify the impact of the offering, be that a sales offering or service offering.

6:33And that takes a lot of different shapes. It could be something as simple as drafting up an email using AI. It could be something as summarizing Sasha as a Bose user using AI. It could be a various amount of things, but Salesforce has always been a tool that's tried to be as declarative as possible so that we're not beholden to having an IT team having to build something out in Salesforce. We try to ship out things that are as declarative as possible. I hope that's very

7:02Sacha: interesting. And you mentioned that customers' expectation really changed over the years and Salesforce is capitalizing on this. I'm curious to know, in your opinion, what influenced or what made customers' expectation change so much?

7:22Guest: That's a very good question. Yeah, that's a tough one. I'm not sure, to be honest with you. I don't know if it's because there are more customer complaints than before. I don't know. It was kind of like the chicken and egg story. I'm not sure what came first. But especially as Salesforce, there's a correlation between how quickly you resolve a customer's problem and the amount of revenue generated from that customer. Because service, as an example, is a second chance to approach a impression. Does that hopefully mean that you're unlocking more sales from the same customer? So, one, it could be reactive where there was a massive influx of service cases, as an example. Two, it could be proactive that, hey, this can be an opportunity to improve the amount of revenue we can get from each customer. But I think, you know, because there's like, it could also be that the demand for a tool like this could just be because technology improved and therefore a tool like this could be created.

8:27Just from a relational database perspective, like the kind of different databases you have, the kind of throughput you can get from that. We've unlocked more abilities to do things as humans interacting with these interfaces. So, it could also be that. It's just that technology unlocked the ability for us to do something we couldn't do before.

8:43Sacha: Yeah, that's a good take. I think it's a combination of, as you said, technology. But also, that's again, that's my opinion. Also, like the generation growing. For example, my generation, we grew up with, you know, technology in our hands, right? iPads, iPhones and everything. Everything at our fingertips. And so, our expectations are always getting bigger and bigger because you can have access to even Google, you know, instantly. So, maybe that ties into, you know, with technology to answer the question.

9:12Guest: Very much so. I agree with that.

9:15Sacha: And then, what are some key AI features in Salesforce that maybe users might not, you know, be aware of at Salesforce?

9:24Guest: Yeah. So, if you were to Google Salesforce, a couple of things that will come up are things like Prompt Builder and Copilot. These are our declarative tools that customers get access to when they purchase any of our AI solutions. Prompt Builder is a way for you to pre-build a prompt for the most common use cases and then print the results of that prompt in whichever field you like. And let me explain what I mean by that. For example, Sasha, you are a customer of mine and I have to reach out to 15 different Sashas every day for different use cases. Can I get my Prompt Builder to essentially generate a human sounding email for me, populating data from Salesforce about you, such as your previous history with purchases, requests about, you know, product changes, whatever, and surface up an email? That's one example. Copilot is essentially an assistant. It's an agent that resides in the same pane of glass as Salesforce for you to ask natural language questions such as, hey, what is the likelihood that Sasha is going to churn? What do you think about me asking Sasha about this particular product?

10:27Are there products that Sasha hasn't bought before? Like if he's already got a really good pair of headphones, does he benefit from maybe buying a really good set of speakers for his home entertainment system or an amplifier, you know, things like that. So those are just some of the example use cases. But there's something that customers haven't learned about just yet with Salesforce is that we actually have our own vector database as well. And what that means is that you can have unstructured data be fed into the models in a safe and secure manner to make sense of Sasha at a deeper level so that Sasha can engage us better.

10:59Sacha: Yeah, and I think that's really powerful. Could you maybe give an example of a successful implementation of a vector database in Salesforce or any other example, maybe?

11:09Guest: Yeah, absolutely. So without actually sharing the name of the customer for their privacy, we have an example of a customer that has very, very long sales cycles and they're in the mining business. So they'll often get RFIs to bid on a particular mine because the job is done and now it has to ecologically be restored back to its original state or as close to that as possible. So they get multiple RFIs from the government and various sources and a lot of the data around those RFIs resides in like Excel spreadsheets. So previous RFIs were submitted and those were archived in like an S3 bucket, right? If you've been a part of RFI cycles, which I wouldn't be surprised you have, it's a painstaking process. A lot of administrative questions, a lot of redundant questions. So we were able to unlock access to all that wealth of information in those previous RFIs using RAG because we're feeding that data back into the vector database so that when a new RFI comes in, it automatically looks at the questions in this new RFI and finds the closest match to the answers from previous RFIs.

12:12So something that used to take 175 days to fill out now takes less than 30. Massive, massive improvements in value. Therefore, they're able to achieve better business outcomes. They're able to win more bids. And it's so much more helpful for them to expand their business this way because we're not eliminating the need for the human. We're making their jobs simpler by taking over the redundancies, the boring tasks, right?

12:37Sacha: Yeah, and that's a powerful ending to your sentence here. And it's funny because I recently listened to Andrew NG. I'm not sure how to pronounce that. And he's saying that AI shouldn't replace a job, but rather a task from a job to make it easier for the worker. And I think that's really cool to see that at play in Salesforce.

12:59Guest: Yeah, 100%. I would love for you to approach Andrew NG for a session with you on this podcast. He's an incredible figure in AI.

13:07Sacha: Yeah, truly. And I'm curious to know how does Salesforce AI capabilities, because you've mentioned quite some, compare to other major players in the industry, if you don't mind sharing.

13:19Guest: Yeah. So I think with Salesforce, they've taken some key concepts around AI and applied them to any product that they've released. One is that AI should not come at the cost of trust to the customer. We take that very seriously. So we built out our own trust layer, as we call it, and essentially a firewall between any LLM and Salesforce. And it has a couple of main tasks. One is to make sure that the LLM does not get trained on customer data. So that retention stays on the Salesforce side boundary. So if I have all this information about Sasha that I'm feeding into the LLM to get me a better result of an email, that data has to stay retained in Salesforce. That's just one example. The second is having an audit trail of all the different queries that have been conducted, things such as also toxicity detection. Are we applying AI for ethical uses, or are there times that we're maybe asking the wrong questions? Things like that. So that's one of our advantages in Salesforce is that trust truly is one of the values. It sounds like a buzzword, but it's been the value of Salesforce since its inception.

14:21And with AI, it's gotten even more important. The second thing is I think we are being very careful with when we are replacing a task versus replacing a human, to your earlier point. In any AI application in Salesforce right now, there's always a human in the loop. Because we're dealing with B2B customers. It's not like a Spotify recommendation. This is really serious shit, right? Like, am I recommending a promotion to a customer that probably isn't entitled to one? Could that be bleeding dollars out of the business? Could that be creating customer satisfaction issues? Could that be creating trust gaps? So the human in the loop element is important here at Salesforce. I'd be curious to see how that evolves. But right now, I think those are two key things. The third thing is AI is just augmenting the things that already exist in Salesforce. So a relational database with a UI layer, the ability to customize Salesforce to your business, industry solutions, automations, really, really important automations, right?

15:21So it just it gives every one of our products a bit of a facelift. And that's kind of where Salesforce is positioned right now. We're up against our competitors without naming them all day, every day. But we stand to learn from them as much as they stand to learn from us. So at Salesforce, we have the culture of learning. And that means that we are encouraged to attend conferences of our competitors. Truly understand without being a Salesforce employee, what that experience looks like and be humble in learning about what that looks like. And if it's better, take the L and understand how we can improve, right?

15:49Sacha: That's really cool. So responsible AI, human in the loop. I think those are emerging trends. They've been emerging. And I think they're really, really important as well. So that's great to see that Salesforce is actually working towards that at its core. So that's really fantastic. And talking about trends, I'm very curious to see what are some emerging trends in AI that you think personally will have a significant impact in the next five, 10 years.

16:16Guest: I actually think, man, I truly believe RAG is really, really interesting. You're probably already ahead of me on this, but I've heard about the emergence of GraphRAG, which is actually in my head. I dig deep into this over a weekend and just read some papers about it. It sounds to me like GraphQL meets RAG. I could be totally wrong. But I think RAG is a bit of a misunderstood piece of technology within AI. So I think that's probably one that's going to stick around for a long time. But I actually think that we're going to learn a lot about us as humans, that we actually maybe resort to structure more than unstructured. What I mean by that is sometimes we'll have customers ask us, hey, can I just interact with my CRM in an unstructured way? Where I just use my voice to ask it to do things, and it just does. And there was a few articles about is AI, does AI mean the death of structured relational databases? I don't think so. I think we desire structure and we desire the visibility to the structure.

17:17So to answer your question, I think RAG is going to play an important part in helping us decipher unstructured data from the past, but also help us desire structure in the future while leaving some flexibility in things that may not be as structured. One thing I'll add is the AWS CTO did a really good keynote on this, and he used the matrix analogy in his video. I would recommend anybody check that out. It was from their reInvent conference. But anyway, to answer your question, that's why I think RAG is the underdog.

17:49Sacha: I fully agree, actually, because it's funny how, I don't know, like maybe five years ago, everyone was saying data is the new fuel, like the new gold, whatever you want to call it. And it seems like now this discussion has shifted to like AI being the central discussion. But I think you couldn't have good AI without proper data, right? And I think that's also why RAG will stay and become more and more important because fundamentally, it's really on your data. If you don't have good data, then you won't have good AI. And I think that's something that Salesforce really understood because they're using enterprise data to really amplify this AI superpower, if you want to call it this way. So I fully agree.

18:36Guest: One thing I was going to add there is the biggest winner out of this whole AI boom, has been technology that's decades old, which is like machine learning. So customers that weren't taking machine learning seriously a few years ago are using AI as a scapegoat to do something that's been around for a long time. To just go back to a good old regression model, you know, like just good old, like, what is the likelihood for my opportunity to close? What's the likelihood for this opportunity to come in this month, as an example, right? They may call it AI and that's fine, but really it's just machine learning. It's just simple math. So I'm excited about that as well. Just like a resurgence of ML back in the mainstream.

19:13Sacha: That's true. That's true. That's something I've seen a lot as well. So you've talked a little bit about it, but how do you see AI transforming maybe like the workplace or even different industries in the near future?

19:25Guest: Yeah, I think some industries more than others. I think in the healthcare and life sciences industry is going to be disrupted hardcore for good reason. Like a radiologist may be able to use AI to detect cancer much, much earlier than they used to before. They'll still be a human in the loop, but they'll be able to get some assistance from AI. I think about things like law firms and lawyers. I feel like their world is going to be disrupted pretty hardcore. Let's assume that there's just a regular market economy at place and that there's no tampering within that through the invisible hand. I think the law firms will start to see quite a difference in the way they are utilized. From a CRM standpoint, I think from a technology perspective, we're flexible in that and we are exploring ways to meet our customers' industries and find out what's important to them. But I think, to be honest with you, I would love to see like an AI email spam filter so that we go back to just drafting real emails. And this is kind of odd as me saying as a Salesforce solution engineer, but I think it will help the people that desire human connection stand out versus the ones that may have taken a shortcut.

20:41And I hope that once the AI boom kind of stabilizes, we get more empathetic because I think we're jumping to solutions before listening. And I find myself doing this when a customer is asking me a question. I got ChatGPT open and I'm already trying to understand the answer to it rather than trying to understand the question in much more depth than I used to.

21:03Sacha: And I don't want to take a different tangent, but on that topic, I've noticed or observed that the more we progress with technology, the more we go back to community, human interaction. And it's pretty funny to me how these two are supposedly the opposite, right? It's technology. We're making such like progress and with AI, ChatGPT, whatever, you know. But yet we're going back to wanting more human interaction. And I'm very curious to see or to hear what you think about this.

21:36Guest: Oh, I couldn't agree more. You're absolutely spot on about that. I see this just, you know, in my own friends. But I truly believe, like, I would love to see some data around, has there been a resurgence of, let's say, the amount of campers, as an example? How many more people are going off the grid as we start to see the spike in technology? Is there a rise in that? If so, then you're absolutely right. We desire more human interactions, more community feel, less technology as technology gets simpler. Kind of strange, isn't it?

22:10Sacha: Very strange. And I would like to give you the floor to, you know, say whatever last word you want to our audience.

22:18Guest: Well, no, thanks. Thanks for inviting me to this, Sash. Parting word would just be, you know, think about which parts of your life you want automated in AI in two years, not just what you want automated tomorrow. And think about the psychological impact every time you want to use AI versus just intuition or using human interaction. And desire to be perfectly imperfect rather than be perfect.

22:45Sacha: That's beautiful. Thank you so much, Jag. It was truly an honor to have you with me today.

22:50Guest: Thanks for the invite, Sash.