AI Driven Digital Customer Success in App Form w/ Dickey Singh of Cast.app | Episode 040

Dickey Singh, CEO of Cast joins us this week for a bonus episode in which we discover his fascinating journey from a long line of engineering leadership roles to today, where he's leading the charge of AI in CS. 

We discuss at length what he and his team at Cast.app are doing to help revolutionize digital CS and what the future of all of this technology looks like.

Topics covered include:

• Dickey’s long history with AI and his journey from an engineering background into serving customer success today
• He was ‘there’ at the inception of Customer Success
• How Cast.app  helps its customers achieve 30x ROI via providing digital CS
• 1:Many vs. 1:1 vs. Many:Many in Digital CS
• Getting persona-specific in your outreach, but don’t go by titles and go by responsibilities, interests and user profiles
• Digital provides efficiency and automation to drive efficiency for CSMs
• Executive personas are likely not going to be in the app and a lot of C-level execs prefer text messages
• SMS is a powerful way to reach certain personas and shouldn’t be as frowned upon as it is
• The tendency for AI to hallucinate and how it needs to be verified by the human
• Current and future state of AI in CS and how adopters stand to gain massive efficiencies. More and more useful AI tools are coming.
• Prompting won’t be as relevant in the future as it currently is

Enjoy! I know I sure did...

Dickey's LinkedIn: https://www.linkedin.com/in/dickey/
Cast: https://cast.app/
Pure Storage ROI with Cast: https://cast.app/purestorage-roi

++++++++++++++++++

Transcript

Speaker 1: 0:03

Talk about numbers, talk about, like you know right things that matter to him. Every user at the same account should get a separate subject, separate body, separate email, because you're, at the end of the day, you're trying to communicate as a digital CSM to them, not as a customer marketing expert. We want to send emails like how a human would write hey, I'm concerned about this. Your number dropped by XYZ. Do you want to talk to me about this? Put that as part of the subject.

Speaker 2: 0:29

And, once again, welcome to the digital customer success podcast with me, alex Trokovich. So glad you could join us here today and every week as I seek out and interview leaders and practitioners who are innovating and building great scaled CS programs. My goal is to share what I've learned and to bring you along with me for the ride so that you get the insights that you need to build and evolve your own digital CS program. If you'd like more info, want to get in touch or sign up for the latest updates, go to digital customer successcom. And for now, let's get started.

Speaker 2: 1:03

Hello and welcome back to the digital customer success podcast. Is so great to have you back for episode 40. I can't believe I'm saying that, but it has been 40 episodes since I first said hello to you, and today I'm bringing you an off cycle extra episode for the week with none other than Dickie Singh of castapp. You will likely have heard of Dickie before because he's very active in CS circles and is CEO of cast app, which is such a cool platform because it is one of a few emerging AI based and AI driven CS platforms out there and it's essentially digital customer success in app form. So we have a lot of great conversation about digital CS, but also personalized motions and the future of AI current state of AI such really great and timely information in this episode that I just couldn't wait to share with you. So please enjoy this conversation with Dickie Singh of cast.

Speaker 1: 2:13

Just age myself.

Speaker 2: 2:16

Same here man. Yeah, I remember very distinctly sending my first email. I was, you know, I don't know, I guess I must have been in high school maybe and I remember sending it and calling my friend that I'd emailed and say hey, did you get it?

Speaker 1: 2:40

Nothing is changed. Now you were texting. Hey did you get my email? Because you sure as shit ain't reading it.

Speaker 2: 2:49

It's funny, well, dickie. Hey, I really appreciate you coming on the show. It's a pleasure to have you. I know we had to reschedule a couple of times on my end, I think, once on your end, yeah but we made it happen and I'm really excited because, you know, not only are you doing awesome things for the CS community, but you have just a crazy background as well and I want to dig into that a little bit. But welcome to the show. Thank you so much.

Speaker 2: 3:21

Yeah for sure. So, speaking of background, I do want to kind of maybe rewind the clock a little bit. You spent some time at MIT. From the looks of it, you did some cool stuff at MIT.

Speaker 1: 3:37

Yeah, we did I don't know like an AI course over there before all this new stuff that came out and learned a lot, right. I mean, like we wanted to like use AI everywhere.

Speaker 1: 3:48

So yeah yeah, I'll tell you more about it. Like you know, right now we are using AI as an assistant, right? So then we're talking about using AI as a peer. Then we're talking about like using AI as a manager that manages me. And then there were even like things like organizations, right, like the AI would run the organization and it'll just give you tasks, and there's like distributed organizations and also. So we were very futuristic and we thought, like you know, this is 30 years in the making. Then, openly, I came out and everything like a six months, seven months.

Speaker 2: 4:21

Yeah, crazy, crazy. And you know, everybody kind of thinks that the open AI thing is like, it's like the first thing that is AI, but it's. I mean there's a long history there that has led up to this event in history.

Speaker 1: 4:37

We were using OpenAI, obviously before it became popular, and other will use like four or five different services models, and we ended up like training OpenAI with a whole bunch of stuff, right? Like we said, like you know, our CSMs don't know how to write SQL. Why can't they just describe it in natural language? And it will here. It'll connect with Gain Site, it'll connect with Salesforce, it'll connect with Snowflake and write a query across platforms and the CSM just has to look at the results of the digital CSM in our case, as you just look at the results, if it looks fine, click a button and it will be used to gather data, right? So? So we ended up doing that. We had to train it quite a bit, but now it's available to everyone because, you know, I guess we didn't have business accounts at that time, right?

Speaker 2: 5:20

So early on.

Speaker 1: 5:22

So whatever you're doing, you're able to do everyone.

Speaker 2: 5:26

Yeah, exactly, exactly. Well, so tell me a little bit about your, I guess, what led you up to? I mean, today you're, you know, very well known in CS circles because of CastApp that you know you're founder of, been a like for, for, but you have this crazy rich history of engineering and and and product leadership across all kinds of segments, right, yeah, and I'm curious to get your, your take on kind of what led you down this, this path. What, what kind? What was the evolution to get you into, you know, serving the CS community with software?

Speaker 1: 6:09

Yeah, so. So my background mostly been, like you know, like an SVP of a product or a CTO with six or seven venture backed companies in Silicon Valley. As I said, like I've been here since 95. And before that I was in Virginia and and I had no idea about customer success before I started Castapp and I think that helped in a way as well.

Speaker 1: 6:34

But there are like two stats that I read that kind of were in the back of my mind, right? 75% of the revenue of most companies comes from existing customers and there are like eight to 10x more tools for pre-sales. So it's like you know the exact opposite, right? If we you know 70, 73% in sales for 75% in UiPath, like you look at, look at any any company, look at PLG companies or SLG companies, it's the same thing, right? So, but the number of tools we sell to pre-sales or to sales people is like 10x more.

Speaker 1: 7:10

So I always wanted to like understand it more. So you know, you know what I did is like I bought every book on CS on and on post sales, like you know, read them cover to cover, made yellow marks all throughout, and I tend to like read the books again and again, like you know, because sometimes they oh, I missed this or something. No, that's absolutely wrong, that doesn't work. That's like he's just the author is just selling the company or just selling you know his services, right? So I always kind of thought that way.

Speaker 1: 7:42

But then something interesting happened. I was just having lunch with my one of my ex CEOs at customer sat where we kind of mostly served like Apple, yahoo, salesforce, you know BA, weblogic and you know AMD, those kind of companies and he mentioned something someone about, like Mary Alexander, and I said, oh, she was a board member. As he said yes, so I said like I want to meet her and I had lunch with her and then she told me she formulated or she pioneered or coined is the right word customer success in 1990 for.

Speaker 2: 8:17

Vented.

Speaker 1: 8:18

And you know, and I read all the blog posts from everyone like, oh, Salesforce this is. I didn't believe her to be honest, right, yeah, yeah, yeah. Then she followed up as connected me to another person she and he was her indirect report like maybe two, two, three levels down or something, and he shared with me his customer success manager card. Right, he showed it to me. Like it was, like you know, in a bad shape, yellow and all that stuff she's telling the truth.

Speaker 1: 8:47

So basically, so I got together with like a few executives and I did. What did I do? We did a interview for her, with her, and we kind of actually posted on our cast on a website, castapp slash articles you have to find it it's pretty interesting is even in her mind it was a way to add revenue, not just service the customer. It's like full circle. After so many years now it's going back to talking about revenue. Even if you're not selling or upselling, you're influencing the sale. So that thing was pretty. To be honest, that's how I got to know CS. From her original experience, I always thought it's about doing better for the customers and doing better for the teams.

Speaker 2: 9:37

Yeah, that's a cool genesis. I like that. We literally you started with the genesis of CS. I think a lot of people work backwards into that, but yeah, that's really cool. Well, so fast forward to today. Castapp, pretty well known. The reason why I'm crazy excited to speak with you is because, essentially, what you're doing is digital customer success in app form, which is very just like meta and cool, but like, and so this is a question that I ask all of my guests which is this notion of what is digital customer success? Because it varies from absolutely everybody I talk to.

Speaker 1: 10:22

Yeah, that's an interesting question. I think it's whatever the company is selling. Is digital CS in my mind, like that's how most people think of it. But I mean it could be webinar software, it could be like community software, even email marketing. I guess everyone's jumping into the bandwagon. But seriously, I think, to me digital CS is, I guess, using technology to help customers and even the customer success team members to grow and preserve their revenue.

Speaker 1: 10:56

But, here's a kicker with a magnitude higher speed, accuracy, reliability, consistency and scalability. And that magnitude higher mean like 10x, 20x higher.

Speaker 1: 11:06

And that's what automation and AI really get you is like our customers have, like I'll share with you later. I guess like 30x ROI on some of the products. I mean we take the credit, but at the end of the day, is like combining automation and AI and making sure that your AI does not hallucinate. We spent a lot of time doing that, but we'll talk about that as well.

Speaker 1: 11:28

I personally see digital CSS three things right. One is the very simple one too many even like a blog post or infographic or even a webinar or a webcast that you do. You do it once you share with multiple people. This podcast is two people talking and sharing it with multiple people. And the second is what I kind of started calling many to many like early on, and there are just two things in there. Right, one of them is multiple people helping multiple people, like communities. And the second thing that's becoming really popular these days is customer and customer success manager, collaborative wikis, in other words, right, they would write some post, the customer will comment on it, the other CSS will comment on it and it ended up becoming like a collaborative article. So that is many to many.

Speaker 1: 12:21

The third is what we are doing, which is castaf, which is going back to the roots of customer success, which is like one-on-one personalized digital experiences for customers, but one-on-one. In other words, if I'm talking to a C level executive, my tone changes, my conversation changes. When I'm talking to a CFO, I'm a little bit more professional. I'm talking about ROI. When I'm talking to a CCO, I'm talking about like ROV or things like that. But when I'm talking to a user, I'm talking about hey, learn this, do this, watch this video and surfacing and making summaries of those blog posts, and so we're showing it to them. That's what we do. And then also, when you talk to a power user, don't teach them how to click the mouse or anything.

Speaker 1: 13:07

Like hey, you have too many users in your segment. You should sub-segment them and don't believe in a single segmentation model. Like you know, have several segmentation models. It's okay, gaston, I can handle it right. So those are the kind of things that we talk about.

Speaker 2: 13:23

Yeah, I love that and it is really kind of one of those kind of golden opportunities in digital is to get to the point of sophistication where you can be personas, and it's hard. It's hard to get to. You know, you got to know who your users are and you got to have some data to support that and all that. But once you do it, it's so powerful.

Speaker 1: 13:53

Yeah, I think you personas the most important thing you can talk about, like when we were talking to people. But one advice that I figured out is like, don't go by titles, and the reason I say that is in an SVP at Salesforce may not have that much authority, but like a head of customer success is a very small company would have a lot of authority. So it's a different thing. Don't go by titles, go by responsibilities, go by roles, go by user profiles, go by interests, and then we use all that data to figure out behavior. We have so much AI built in, even to our player kind of thing, so if we notice that someone always skips over support slides or something, we move that away that they can click on and access it, for example.

Speaker 2: 14:38

Yeah, so making it intelligent and really specific to that person. That's super important. In a previous episode I had a really great conversation with Dan Ennis. I don't know if you're familiar with him, but he's been doing some pretty amazing work around data analysis of specific personas, where he's essentially building a data profile of what an admin might be versus what an end user might be, versus what an exec might be or whatever, and then applies that across the user segments to where he doesn't necessarily need to ask them or know for sure what they are, but based on their usage and their behaviors and things like that can get a pretty good sense for what the persona is, and I love that level of sophistication. It sounds like y'all are cracking that very similar thing.

Speaker 1: 15:36

Yeah, yeah, absolutely. And the only thing I would say is don't go by titles, go by responsibilities, like a CEO of a startup would do support and customer success and bring the rights for the team.

Speaker 2: 15:49

From mopping the floors to signing the checks.

Speaker 1: 15:52

Yes, there you go.

Speaker 2: 15:55

Probably at the same time too Good. You've got two minutes. So I want to get a little bit deeper into CAST because I don't like to kind of I hate having founders on and then basically tell them what their app does. I'd much rather hear it from the founders themselves and kind of let you kind of speak about not only the genesis of it but, at its core, what are you solving for?

Speaker 1: 16:28

Yeah, what we are kind of solving is we only see two major problems for every B2B company. It's a huge thing. One is like OK, before I say that there are two or three types of companies Companies that are very simple products, they don't need CSMs, they'll never need CSMs. It's a very simple idea. They will just build it into the product, and these are like young PLG companies. And then there is the whole mass of B2B companies and SaaS companies which kind of rely on CSMs and account managers and subject matter experts to provide adoption and usage and sales and upsells to customers. So that's what we're going to focus on. And the last category is companies that take like nine months to deploy and they have an army of professional services, technical services and all those kind of things. So we are kind of focused on the middle, which is luckily, 95% of the 5 million B2B users, customers right there. Right, 5 million B2B companies. So we're trying to help over there.

Speaker 1: 17:44

And the two things we realized is the CSMs are and account managers, each have too many accounts and, the other hand, not every account has a CSM. So we are just trying to solve those two problems and the way we did it is we think it's very unique, but everyone has his own opinion. So, in the traditional way of doing customer success, you put a customer success manager in between the customer and the playbooks and the data, and you need a CSP tool, you need CRM tools, support tools. So what we did is what if we could elevate that CSM, do a more of a consulting role and maybe so he or she can focus on relationships with customers and new challenges? And, in his place, bring in a digital customer success manager or an AI-driven digital customer success manager, in our case who executes playbooks, digitizes playbooks, executes playbooks and communicates with the customer in the manner that is approved by salespeople or the VP of sales, is approved by the CCO and uses automated benchmarking to upsell and cross-sell and uses things like hey, customers, your size, use these four features. You're only using three. Do you want to talk to your CSM, scott Adams, to see if they want to learn more about this? Or same thing for sales right, when you use competitive benchmarking. So that kind of comes for free with our product. So what we are solving?

Speaker 1: 19:28

Just two simple use cases a digital business review with a purpose. What good is a digital business review when you cannot upsell, cross-sell or recommend products, right. So, and then advanced onboarding, which includes, like pre-boarding prospects, onboarding accounts, onboarding users, onboarding different personas differently, re-onboarding a previously bundled product when a new hire happens at your customer, onboarding the users again when the account has already been onboarded. So the more complex the onboarding. That's where we kind of come in. So two simple use cases, like a digital business reviews could that grow and preserve your revenue and accelerating value realization through re-boarding, onboarding, re-boarding and even off-boarding, although we don't have a single customer that's using our off-boarding part?

Speaker 2: 20:25

I love that so much because it speaks to what digital CS has kind of evolved to over time, whereas and regular listeners will have heard me talk about this before but, like a few years ago, the notion of digital CS is okay, we need this thing, a bunch of customers and that's definitely evolved into what it is today, which is to say, okay, yeah, we're doing a lot of customer facing things, but a big part of it is automating things so that your CSMs can be more strategic, provide more value and less time doing the mundane things sending the same emails, doing those kinds of things. And I love that you're fitting right into that level of kind of sophistication and automation to where it is removing the burden, so that your CS, I hate saying, can do more with less but can be more valuable with less, I suppose.

Speaker 1: 21:30

Yeah, people are sort of hating that term, like doing more with less, but I guess I guess I can see it's like built in right, so yeah, absolutely.

Speaker 1: 21:39

And you may have heard of that term like in a death by campaigns, right. What that really means is like you have to write your own campaigns and the campaigns that the customer marketing team wrote versus the campaigns that the CSM is putting together or versus the campaigns that the marketing. They don't talk to each other. Even your four campaigns don't talk to each other. Now, the biggest problem with such environments is like, if the campaigns don't talk to each other, you're sending too many emails, too many things, and if you upsell or cross-sell in an email, you're going to be banned as a spammer, right, Obviously, because you know, and then you get like sort of getting really poor results. So the trick over there is to not upsell and cross-sell an email, but like take them into an environment, whether which is in-app or out-of-app, which is email or text, so the executive does not have to ever log in, right, yeah, If the executive has to remember how to log in or you're not going to be able to reach a CFO or a Chief Administrative Officer, I'm trying to say is like non-linow business executives who are very influential, whether to keep your product or not keep your product, If they don't know what ROI or ROV.

Speaker 1: 22:48

They're not going to continue that. So that's the underlying thing. Like you know, instead of asking everyone to come into your product just so that you can claim like higher monthly active users or daily active users, what if you could take the product to where the user is? If they send an email or text, reach them over there. If they log in like Power Users and Users, show them in-app.

Speaker 2: 23:11

Yeah, it's interesting too that executive persona. Nine times out of 10, they're not going to be in the app at all, like they're not your users, they're not your intended kind of audience for regular users, so why expect them to log into the app just to kind of see how things are going? And one thing you hit on that I think is something that is very often overlooked is text messaging and SMS. I don't think a lot of people are doing that because I think there's an automatic assumption that it's kind of creepy or, you know, it's invasive or whatnot. My argument has always been that if someone wants to be communicated via text, or if somebody wants to be communicated via email or whatever it is or Slack message, like, we should give our customers that option so that if they want to be communicated with text, they can be. But I think that's it for a lot of people.

Speaker 1: 24:15

Yeah, you're absolutely right. So one distinction that we like to make is let's say you use HubSpot or so for email marketing, you don't have a relationship with the people that you're sending your emails to. When you are talking to a customer, you have a relationship with the people. They're like kind of opted in in your terms and conditions that you can communicate with them. So use transactional emails and that's what we do, and and honestly God, we get like minimum of 2.1x to 4x the industry average when we send out conversations. Like you know we can. Hp, you know they use CSP platforms but send emails to us. Their storage, aruba, same thing, right? So the interesting thing is, in the presentations or digital hubs that we generate and send, we ask them would you like to opt in for getting text? So we have to do an opt in. So what we notice is like C level execs prefer text, whereas operators and everything prefer email and users and power users prefer to just log in into the product or so.

Speaker 2: 25:26

It's like a no brainer, but there's a huge aversion to that. I don't really get that, but I think it's extremely powerful because your executive, they've got their phone like all the time right. What do you recommend when folks are, you know, setting up the system and doing those kinds of things? Do you make recommendations on what they should send via SMS versus email?

Speaker 1: 25:53

Yeah, you let your end customers choose SMS, but you reach out to executives over email, but give them an option to switch to SMS. But then the campaign should also change, as I was mentioning earlier. Right, like our camp, you know, you don't have to create multiple cameras because we have a generative platform. So not only the presentations are generated, the digital hubs are generated, but even the campaigns are generated, right? So if I take one 30 second departure, why is that helpful? Because if somebody has already done steps eight and 12, the campaign won't get stuck, it will overlook that. Tell them to do steps one to seven and then, as of two, step nine and then it move on. Right, because it's a generative platform.

Speaker 1: 26:35

Since it's a generative, don't write like really long emails, put like really short emails. Give them a single call to action button, right. Or, if you have several call to actions, they should point to the same result, which is like the presentation or view, your digital hub or so. And then, if you're sending it by text, should be like one or two lines, right? You know, we have come up like we call it, like anti patterns of emails. Like you know, everyone puts the name of the customer on the top in the subject and that does this.

Speaker 1: 27:05

No, talk about numbers. Talk about, like you know right, things that matter to him. Every user at the same account should get a separate subject, separate body, separate email, because at the end of the day, you're trying to communicate as a digital CSN to them, not as a customer marketing. We support customer marketing also in our product, but we want to send emails like how a human would write hey, I'm concerned about this. Your number dropped by XYZ. Do you want to talk to me about this? Put that as part of the subject, right. So, and again, you can do that with generative platforms. It's very hard to set it up when you are kind of creating campaigns separately.

Speaker 2: 27:43

Yeah, yeah, exactly, exactly, are there? Are there some like particular, like success stories or really cool like use cases that you've come across? Yeah, and customers that are using it particularly well? Yeah, absolutely.

Speaker 1: 27:58

I mean, yeah, there are quite a few. Right, there's a paper on our website about pure storage. This is the first division that signed up. They are making over $400,000 a quarter and they're attributing it to digital CSMs. Wow. And, and another division of them forget, I think they do like Six Sigma training for customers, or so they save over 150 hours per account by switching from a cell service LMS to gas personalized learning.

Speaker 1: 28:40

Now, the difference in LMS is like, you know, yeah, you have a 30 minute video. Or you watch that, but or we can generate like a four minute conversation. Five minute conversation that goes over and teaches you the five vies of Six Sigma, but also teaches you how to use, like, fish boning and all those kind of things. In case you have forgotten why? Because the prerequisites you need to know fish boning before you do this, right? So the presentation video that goes out is custom created for Alex. Why? Because we know that Alex has taken these four courses. He got three A's and one D, so we'll talk more about the thing that he got. Yeah, right, it's like personalized, because we know this, we know, and it's much shorter, right? Like, if what is four or five minutes, is there an option to look into that.

Speaker 2: 29:32

So those are some of the yeah it's if that's a cool combination of kind of the concept of micro learning and and AI generated kind of content. Curation is essentially what it is. Yeah, yeah.

Speaker 1: 29:45

Yet another example this customer route this and I tell this to all my customers because I they've given me permission to share some of their analytics they get like four, x, the other people as far as like success metrics are concerned, and they have a very simple philosophy Less is more. They don't want to tell more things to a customer. They show like two or three slides, but they let them ask questions. They let them. So I don't know whether you're aware or not that our product trains on every tech stack and your products automatically. So what that? What that really means is like when you are giving a presentation, you usually show seven to 10 slides, but on your eighth slide, the CCO has 20 questions, right? So what normally people do in a QBR or something, they create like a 50 page deck and have a big appendix and all that stuff, right, but that's not the right thing. You just keep it very small, but when the people ask questions, ask them, answer their questions live using an agent, right? So that's what we do.

Speaker 1: 30:50

It's quite interesting. It was very well received by HP Pure Storage, aruba, v-com, route, this and, yeah, I mean HP shared something that engaged customers by 90% more often and spent 60% more for transaction. So I mean, yeah, they have been using digital and our product for a long time. They were one of our design customers and yet another customer which I should be able to mention. Early next month they showed us that a CSM can handle 2.17x the number of accounts across segments and I have to read this while improving the CSM NPS by 16% and improving the onboarding CES by 21%.

Speaker 1: 31:38

So, at the same time, right. So basically what I'm saying is the CSM can handle 2.17x the number of accounts across segments and the CSM NPS improved by 16% and the onboarding exports CES in a customer effort score improved by 21%.

Speaker 2: 31:54

To me, what this boils down to is finding those key moments and just really digging in on those and being hyper focused on the outcome and the metric behind the outcome. Because, again to the earlier SMS example, right, if you can text your executive and say, hey, your team is doing this well, your team isn't doing this well, this is your ROI, etc. Etc. In like one or two lines, I mean, how impactful is that versus the 20 page deck that we talked about? That has the huge, but why just yeah?

Speaker 1: 32:39

and that's the difference. Why just inform the user right, tell them what to do. In other words, hey, great job on this, your team is doing really good. This is the next thing that you should focus on. Really bad job on this. You should consider going to this training. So the purpose is not like informing. The purpose is like an action that was generated on the fly. You know we take credit, but it's the AI doing it right. So, on the fly telling them what to do. So never show them negatives. Always use the negative to tell them what to do.

Speaker 1: 33:11

Because, then, and nine out of 10 times you're right. Right, but there may be times you're not right, but still, you gave us a suggestion and you can always say I don't have a suggestion. Do you want to set up like 30 minutes with the CSM? Right, we do that. All the you know, we do that, as I was saying earlier.

Speaker 1: 33:27

Right, ai tends to hallucinate. So what we do is like we figured out with every answer we ask for, how confident are you? You know, the confidence level is less than 0.36. We say we think this is the answer, but we are not sure. But if it is less than 0.26 or something, we say we don't know the answer. But here is the calendar for a CSM. Like when we introduce like these, being able to connect CSM, and people may say, yeah, so I can put this in my database and we, you know everyone can connect to calendar, so it's not no big deal. But this is the reason why we did that is because the AI tends to hallucinate. And then we can bring in the CSM as an end need.

Speaker 2: 34:08

As the human backup? Yeah, yeah, do you feel like you're your newness to customer success and you, coming at this from an outside of CS perspective, has served you well in this? I mean, that's my impression, because you have this other view and you're not bogged down by the traditional sentiment of what a QBR is or whatever. Do you feel like that's served you well in this whole thing?

Speaker 1: 34:45

I hope so, I believe so, I hope so. But you know, see, whatever happened in customer service, like 15 years ago, is happening in customer success now. So in customers, you cannot ask for help without having. When you start typing a question, it starts giving you suggestions right, you go anywhere, but the problem is they are going against a static document, right? In other words, they're going through your health documents and or whatever, like you know, university documents or whatever we call it schoolcastapp. So they're going through that. So what we are doing different from over there is we are reading account information and helping the information.

Speaker 1: 35:25

But that's an aside. But that's the point that I'm making is like, whatever happened in customer service will happen in customer success, which is like more automation and more AI agents answering your question and bringing in an export, which is like the person who did the customer service agent later on, right? So for that to happen, what happened in that industry is they started paying more to customer service agents, right? The humans? Yeah, I call them coffee driven CSMs and AI driven CSMs, right? So that makes it easier for me to explain.

Speaker 2: 35:59

I love it.

Speaker 1: 36:00

Yeah, same thing is going to happen, like the AI driven CSM is going to request the coffee driven CSM to like come in and help her right. So I'll see. We'll see a lot more of that.

Speaker 2: 36:13

Yeah, if you haven't trademarked that already, I would do that immediately.

Speaker 1: 36:19

I should.

Speaker 2: 36:22

Well, I mean speaking of seeing more of that. I think you know, arguably, ai has been around for a while, but we're now getting into a level of usability of gen AI that is obviously unprecedented, and a lot of people are coming to terms with that, some better than others. But what is, in your opinion? What is? What are the next few years look like in terms of the adoption of this technology, and I guess, specifically to CS, like what do you see all this going?

Speaker 1: 37:00

So I mean, we have been noticing this since early October already. Right, like people are not that. Even like we were signing up, a customer just sent the document sign early morning and he said we'll sign as soon as I pick up the kids. So even like security companies are adopting AI, you know. Obviously they ask us tough questions and then we have I have a demo where I put his name and I say who is such and such, right, the CCO's name? And in the past we have tried to answer that no, we cannot answer that because this person is talking to you, you know. So it says like I don't know, do you want to connect with our CSM to answer? So that gives a good way to tell them that people are getting comfortable with this AI thing.

Speaker 1: 37:51

At the end of the day, I don't think so. There are still few people who are resisting, right, like say, hey, human touch and all those kind of things. But I feel it can do more service to you and make your job as a CSM or as a CS leader a lot easier and you can look at like minimum 12X to 40X ROI with the products that you use, instead of using a system of record tool that put data in it and you run reports and then you create campaigns and reach out to people and say nobody's clicking on my campaigns. You know things like that. So things that can learn can really help you take it to the next level. Thank you. If I may, a lot has changed. When I was at the MIT AI thing, we were focused on supervised learning and unsupervised learning. We we even have unsupervised learning in our player because it learns if you always skip over support and it will yeah, to the side right.

Speaker 1: 38:49

But then again Some I supervised learning was a big thing at that time, like we were like, oh, we're waiting for it to like show up. And then we were talking about like one shot learning, a few shot learning, which is like just give it a few samples and it learns. And that's what open AI was like just a year and a half ago. But when we were like training them how to generate SQL from commands or we're Telling it how to have a combination generate code from liquid, which we do that today Also, right, we were doing that.

Speaker 1: 39:18

But now you don't even have to train as part of the prompt. You can give an example here, a couple of examples, just these two or three examples. You know, yes, we want to take credit because we trained it, but at the end of the day, everyone can use those kind of things. Right, few shot learning. So I think that's where the world is headed and this whole opening I Sam Alton fiasco, whether Whether it was because of a GI or whatever the real reason is. Hopefully we'll find out when he writes his book.

Speaker 2: 39:46

But because you know it's coming. Yeah.

Speaker 1: 39:50

Yeah, but yeah, I mean again, they're okay, wall Street is saying one thing, the circle valley saying another thing, but but, at the end of the day, more AI is coming, more useful AI is and I'm not talking about what MIT should teach us like it are using AI as an assistant and eventually appear and eventually Like your manager and eventually, like the, it runs the corporation and just gives you tasks, or yeah, that we are far from that. But at the end of the day, any tool that's available, I mean, yeah, look at, look at the thing no one's hiring. We're the chief of staff, because this open AI, this tool, is your chief of staff.

Speaker 2: 40:28

Yeah, yeah, yeah, absolutely do you. So you know, I think back to kind of Google and early kind of search engines and back Back in the day I mean, we were dating ourselves earlier. Right, I'm giving our early influence, influences of the internet or whatever. But back in the early days you kind of had to know how to search for something Like on Google to to get accurate results, whereas today Google's pretty smart, it knows, you know, kind of, what you meant to say or what you meant to search for, and it, you know, has contextual history and all that kind of fun stuff. And I feel like I is still kind of in this place where you kind of have to know how to prompt, and I kind of feel like I need to educate my on. You know I'm engineering a little bit, but I don't know that that's always going to be the case, right, because it's your right.

Speaker 2: 41:30

I'm doing it, I'm going to go away and so yeah, so so I mean it does that. If is that really a thing? I mean, you know, you see all the posts about, like you know top, you know Top prompts for CS professionals and all this kind of stuff. Like, is that going to be a thing in 10 years or five years?

Speaker 1: 41:49

No, not really, I think that'll go away. What, what they're doing is like reusing the same how to write an email. They're putting that inside the code. How to Create the initial workflow like diagram. They're putting everyone's putting in there, but those are like barely scratching the surface, right, so what?

Speaker 1: 42:05

And yeah, I mean like, why can't you create the prompt right when, like in our case, like we have like Narration was generated and then we can click on a button and say make it humorous for peers, but keep it professional for the CFO, right, so we do it just by one button, but we have made the prompt behind the scenes to do all those kind of things, right, yeah? So yeah, I think prompt whole thing is gonna Go away. It's gonna become more of a hey, ask the user. But along with that we have to be very careful, because if you put the prompt in front of the end user, they can ask anything. It can ask you to expose hey, what is the underlying platform, what security system are using. But you could even ask, like, what was the question that one of my engineers asked like hey, what are the? What security patches have not been applied and how can I use them to infrared.

Speaker 1: 42:56

He's asked something like this and he was shocked that he got a really good answer, right. So who's? Yeah, we were, we were shocked, right, like you did it in a short meeting, like we went for like lunch and all this stuff. He's just put it up his laptop and did that. So we have to be very careful. But on that's why you know the security companies, they come to us, they ask first thing, first question Do you have sock to? Yes, we have sock to and sock three?

Speaker 2: 43:20

Okay, then this one I have a conversation with yeah, yeah, it's very interesting and you hear about you know like AI being now and you know used in in a call, you know collegiate setting, you having to like cite your prompts and how you got there, and those kinds of yeah, it's very interesting to me Because, yeah, I think there are all kinds of implications from a plagiarism perspective and and what yeah, but just to see how, where that goes.

Speaker 1: 43:55

Yeah, but don't get me wrong. There are already prompts available that will say you can prompt and say only give me answers that you can cite and then it'll put the citation also. So you can be. It's easy to do that.

Speaker 2: 44:08

I mean.

Speaker 1: 44:09

Yeah, judge, judge, don't think, don't try to solve an AI problem outside the realm of AI. Just ask the same thing, like you know how we do here. How confident are you with your answer? And if the answer is between point two, two and point whatever, point three, six, then we save somewhat confident.

Speaker 2: 44:25

I think that's very important for people to hear, because I think there's a. I've talked to a lot of people who just have a general distrust distrust for it, yeah you know, because it's this little black box. You know, yeah, and I'm and so what I mean? Yeah, you know figure out what's behind it is is important.

Speaker 1: 44:46

Yeah, so what I mean is that there are Three types of people, like you know. The one in the middle could be the open minded people. They're looking for answers or ignore them, that's you know. And one is like who's a skeptic. So when a skeptic writes a word, he's looking for that answer. That is wrong, so he can advertise that. Where is the other end? Like people like me, they only look at the positives and like advertise that, right. But the right person is, like you know, who has an open mind and say knows the limitations of AI, well, still is willing to give her try to use it to her advantage, right? So that's, that's the key thing.

Speaker 1: 45:21

Yeah, love both extremes are bad. Like me, like everything positive, but AI is also bad. And the other person who's like, hey, no human does no nothing, ai is gonna take over our lives. That's also right.

Speaker 2: 45:33

Yeah, yeah, yeah, balance. It's all about balance.

Speaker 1: 45:38

I could have said that and giving a long story.

Speaker 2: 45:41

It's perfect. Yeah, it's, it's fascinating. I could. I think I could talk about this stuff all day. But you know, the unfortunate thing about the fact that it is eight till the hour is that we are running out of time and that sucks. But I Do have a couple of just kind of round out questions. Yeah, the first one is what? What are you paying attention to? Listening to that informs kind of who you are. It could be business, could be not business.

Speaker 1: 46:14

I'll be honest, I think I'm spending a whale a lot of time on tiktok these days and somehow it had figured, had figured out that I read a lot about entrepreneurship and I kind of you know, look at like AI, hallucination problems and AI, so those three things it talks a lot about. And it also talks to me about shoes, yeah, and sunglasses, because it knows that I collect sunglasses and it knows that I collect shoes, right, or her boots and stuff. So between these five topics I use like tiktok. But then I, when I go for walks, I'm usually have like one airport and I'm listening to books. I tend to listen to the books that I've read before.

Speaker 1: 46:56

Because, I'm trying to look for more things that I missed or I've started to disagree with a lot of stuff that I've re-listened to now, I guess, as I'm evolving as my concept. So I figure out like, okay, they're just saying the norm, for example, their product is for CSMs, so there is no way they are interested in having a high ratio of account to CSN because they charge with CSN, so that's why they're going to say this. So I've become skeptical, since I was just learning from that and I'm saying why are they doing that? So I've started to think like that, which I think people should listen or read books that they have read before. Again after like a three or six month period.

Speaker 2: 47:43

Yeah, I think that's really helped, the fact that you're reading it and then listening to it, because I feel like I know our brain very differently if whether it's auditory input or visual input. So you're actually right, you're probably picking up on stuff just because of that very thing.

Speaker 1: 48:01

Yeah, and what you're describing is and this is one of the research that we did when we started cast. So University of Washington, some biomedicine doctor. He said like if you just use one model, which is like reading a blog post or listening to a podcast, you retain 10 percent of the information. If you add one more, like visuals as well as reading, it goes to 65 percent, and I have the research to actually paid for it. But if you add a third one, it goes to 73 percent. You add a fourth one, 73.5. So it doesn't matter, but minimum two is kind of. That is the reason our avatars have lip movement when they talk. Oh, wow, right.

Speaker 1: 48:42

So you're listening to something you are watching and you see the lip movement and all this. This gave directly from the professor, right, like, but are we very transparent? Like 50 percent of our customers turn the avatar off and 50 percent, like, want to label it, name it. Like I said, hi I'm Amy or hi I'm Mark, depending on the gender. So they do all sorts of things. But you're absolutely right. Sorry for the you know.

Speaker 2: 49:07

No, it's great.

Speaker 1: 49:08

I'll look up the research. Actually, it might be linked on our website also at the bottom. I'll look it up and see.

Speaker 2: 49:13

Well, I mean, you know it's even in editing this, you know these podcast episodes. You know I do edit the video and the audio at this end. I notice that when I'm when I'm looking at it and I've even had this feedback when people are listening on YouTube versus just on their podcast app, they pick up different things Because obviously so much of our communication as humans is visual.

Speaker 1: 49:37

Yeah, so, yeah, I, you know I have CarPlay and recently the podcast app in CarPlay enabled like the 1.5X, 1.7x feed, right. So what I noticed? Like if I listen and 1.7X feed, I'm driving much faster. So I had to like go back to 1.25 and 1.5, because, you know, I'm into sports cars a little bit, but I tend to drive a lot faster when I'm listening to something really fast, right. So it's like that was the reason why I switched it back to 1.5 or you know, versus like 2X.

Speaker 2: 50:11

That's really funny. Apple's probably going to have to update their terms and conditions for that, Just for me just because of this guy who tends to drive faster when he's just taking faster podcasts. Exactly, Exactly. Well, look, I've. I've enjoyed this conversation tremendously. I'm sure people can find you on LinkedIn. Are there other resources or places? People?

Speaker 1: 50:32

Yeah, the eialingtoncom slash in slash deki d-i-c-k-e-y, and the best way to reach me is like my email. So d-i-c-k-e-y-at-castapp, I think those are the two best ways to reach me.

Speaker 2: 50:48

Amazing. Well, again, I appreciate it. Thanks for the time. It's been a pleasure. Thank you for joining me for this episode of the Digital Customer Success Podcast. If you like what we're doing, consider leaving us a review on your podcast platform of choice. It really helps us to grow and to provide value to a broader audience. You can view the Digital Customer Success Definition word map and get more details about the show at digitalcustomersuccesscom. My name is Alex Turchovich. Thanks again for joining and we'll see you next time.