Unlocking AI Innovation: Insights from Dreamforce and MACON

00:03 - Eric Cook (Co-host)

The whole concept if you're listening to this and you're not exactly sure what. O1 Preview, formerly known as Strawberry and then QSTAR, even before that. The concept that we've all been operating under, as it relates to these GPTs and large language models, is these tools have gone out and scoured the internet, the world and the analogy of a librarian that goes to every library on every continent across the entire world and reads every book and consumes all that information and houses it within a large language model, and has developed a prediction engine based off of the GPT framework generative, pre-trained transformer that gives it the ability to say this is what likely is going to be a response, whether it's a written response or an image response or an audio response or a video response. But at no point did any of this technology take into consideration what was going on behind the scenes in the mind of the creator that was writing that book or doing the painting or composing the music. They only have access to the end result, the finished product.

01:08

And so what? Gpt-01 Preview? And I think what this agent force is trying to do is get in the head of the human to try to think about well, why did this book come out the way that it did. And maybe we don't want to get in Stephen King's head, because that might be a really scary place, but what was going on inside of Stephen King's head when he was coming up with this story? How did he approach his storylines, how did he develop his characters? How did he come up with the plot and the twist and the end? Those are the things that, when the model is forced to take its time to be able to reason, to be able to check its own work, Hello listeners and welcome to Banking on Disruption.

02:05 - Fred Cadena (Host)

I'm Fred Cadena. This week, we are mixing things up a little bit. Instead of an interview followed by quick takes, this episode will be one conversation with Josh, eric and I. Last week, josh and I both attended the Dreamforce Salesforce conference in San Francisco and, no surprise, ai was front and center. During his keynote, salesforce CEO Mark Benioff compared the launch of their AI platform, agentforce, to Waymo autonomous cars, highlighting both the game-changing innovation and the inherent risks. Two weeks ago, our other QuickTakes panelist, eric Cook, attended MyCon in Cleveland. From Eric's early report, mycon left attendees buzzing with insights on the rapid acceleration of AI in marketing, while our little pod has never strayed from the topic of AI. On this week's episode, we are going to do a deep dive on all of Josh, eric and my bankified takeaways from these two blockbuster conferences. While you're listening to this podcast, why not take a moment to follow us on LinkedIn at the Banking on Disruption podcast, and on Instagram at at Banking on Disruption? Now sit back and strap in, because our show is coming to you right now. All right, and we're back.

03:22

I am super excited for this week's episode. Rather than having a guest followed by quick takes. I talked with Josh and Eric over the last few days and just really thought it would make sense for us to deep dive and dig into the conferences that the three of us spent most of the last two weeks at. So Eric was at MACON October 10th no, sorry, september 10th, 11th and 12th so two weeks ago and then last week, josh and I were at Dreamforce in San Francisco, and both of these conferences had a significant overlap in just the prevalence of AI and AI-related announcements and focus.

04:07

And really just because of all of this, you know, kind of coming down the pike, I wanted to do a recap with Eric on what he did and I wanted to make sure we covered Dreamforce, and I just think the three of us will have a lot to say collectively about both. And so, eric, I'll start with you, since Macon was first. I'm going to ask you to do a lot in a short bit of time, but can you give like the 50,000 foot? Like you know what was, what was the big? So what's for you?

04:38 - Eric Cook (Co-host)

Well, the the first thing that I found valuable is and I saw a lot of people doing this using some form of AI to capture notes and to help with the digestion of all of the fire hose that we were drinking from for those two and a half days, and I saw a number of people that had Otter and other apps loaded on their phone. I think I've talked about my Plod Notes and it was kind of fun showing that off because I'd pull that little card-shaped device off the back of my phone and people were like what is this magical device that you're talking about? And so I feel really comfortable coming back from the event because there was so much information. There were almost 70 sessions between the keynote main stage, the breakout sessions and the technology demos. There was no way you could be at all of them, and I did get the all access pass, so I've got access to a lot of the main stage items and a few of the other events. As far as their slides go and it was kind of ironic everybody joked that OpenAI waited until the final day, but two hours prior to the conclusion of the event on Thursday, openai dropped its new Strawberry codenamed now O1 Preview, and really sent Paul and Mike into a little bit of a tiz, because Paul got access to it but Mike didn't, and so they went in and did a really quick deep dive to play around with it, have a little bit of conversation. So it was a lot of in the moment, a lot of spontaneity. But the sessions that I attended were amazing, I would say.

06:18

The AI components of the event, certainly custom language models being able to go beyond just what you're getting from a chat, gpt or a cloud. Taking the time to train those models with your own data and being able to educate it with your own content. Whether you start that internally for a help desk or a support or just doing some beta testing but then releasing that information to customers or other use case scenarios where you're really leveraging a large language model type of an environment but with your own custom information. Number of sessions on that. There were a couple of really great sessions on AI efficiencies and the arrival soon of agents to be able to automate things. The arrival soon of agents to be able to automate things.

07:09

latform that connects to over: 1900

08:25

So something a little bit more advanced than a HeyGen that's out there now where it still looks like it's kind of an AI generated goofy that doesn't really look like Eric. But there are studios that are cropping up that are actually for several thousands of dollars. You can go in, you can be recorded with professional grade recording equipment on a professional grade green screen with high level audio and being able to train those and license them and continue to get those better and better and better. So that was kind of interesting, and probably a lot of others, but I'll say this last one and then I'll pause because there's probably more that we can talk about.

09:07

That overlaps with Dreamforce with a click of a button. That human involvement, genuine human experiences, over time are going to become a lot more valuable because people just aren't going to have them. There's going to be so much high quality generated synthetic content that's going to look really good, it's going to sound really good, it's going to be informational, it's going to be free of hallucinations, it's going to be really valid, but the human element is not going to be there, no matter how hard we try. And so things like this podcast where you and I the three of us are all chatting, we're having an interaction. This is not AI generated Conferences and events where you're in person.

09:59 - Josh Matthews (Co-host)

I'm sorry. I'm actually obligated by my code to share that I am actually an AI agent.

10:06 - Eric Cook (Co-host)

Oh, nice, nice. You're doing a really good job with the cigar smoke too, by the way. Thanks, that's coming across.

10:12 - Fred Cadena (Host)

Is that an actual cigar or is it?

10:14 - Josh Matthews (Co-host)

a. Cgi cigar, it's CGI, cgi.

10:17 - Eric Cook (Co-host)

Yeah, yeah. But joking aside, I think it's going to probably like anything else it's going to wear I've already seen people talking about on LinkedIn, poking fun at, Notebook LM. That seems to be the bell of the ball right now that everybody's talking about. But it's becoming really easy to create the synthetic content and it's still going to sound very generic and non-on-brand. If you don't remember that. You've got to put your spin on it. You've got to add your element to it.

10:52

We're not to the point where we can, just as Ron Popeil says, set it and forget it and walk away and think everything's going to be good to go. It's got to have your involvement in it, and I think a lot of people are missing that. Not so much, really, at MAKON, because I think that's a segment that are there that understand that, but the general populace of people that wouldn't even know that a MAKON existed. They're going to see this and go, oh, I can do this, click, click, boom and next thing you know they think they've got something that's great and in reality nobody wants to tell the emperor. They have no clothes and it's going to be not good for some organizations. And so you know, listeners of this podcast hopefully have the benefit of hearing that and reminding you that you have to make sure that you've got human oversight and don't lose fact of the creativity and all the different things that make us as carbon-based life forms yeah, at least at this point, right, human in oversight, right, because, totally, yeah, absolutely, that's where it comes in.

11:57 - Josh Matthews (Co-host)

It's like your, you know, your, your 50k admin, you get. You're like, yeah, make sure, this sounds like me. It's like okay, but come on, you actually have to have your own fingers on the buttons, so to speak.

12:12 - Eric Cook (Co-host)

Yeah, absolutely. So I've got my Plod Notes. I did fire up a Google Doc. I did some live scribing while I was at the conference. Some members of the LinkBanker community actually got access to my notes. I was dropping pictures in of some slides and some other things.

12:30

Just really really good conference. I've already got my ticket for next year coming up in October 25 in Cleveland on the 14th through the 16th. So I will be going back for sure and, yeah, it validated a lot of what I thought was necessary. We had some really good conversations with paul, but even deeper, with mike kaput, their chief content officer. He was, you know, speaker at our global convention last year and I got to know him pretty well. So we become friends and we got a chance for some one-on-one time on the second day and, uh, share just a little bit about what's going on in the WSI world and he was pretty excited. So we're at our convention this week and we're going to be talking about all of the things that are going on embracing it within the organization. So it's going to be pretty cool.

13:18 - Fred Cadena (Host)

So I could ramble on, but yeah, there you go. No, that's great, it's a great intro and I think I mean there's definitely a lot for us to dig in there. Josh, let's start with you next on Dreamforce, and then I'll give some thoughts on Dreamforce and we'll get into it.

13:33 - Josh Matthews (Co-host)

Yeah. So I mean, look, I've got a little caveat here, which is that my experience at Dreamforce tends to be substantially different from a lot of the people who go, because I go to maybe two sessions. I'm not there to learn a lot, I'm there for building connections that exist, creating new connections that didn't exist before and generating some content. And then, to a great extent, it's also like it dawned on me after this past week that when I go to Dreamforce it's really like like going to a hometown I used to live in. You walk down the street and you hear your name called and there wasn't an event or party that I went to, that I didn't buy, purely by accident, run into two or three people that I knew, you know, and and that's always just great from like a community feel standpoint. So I think in that way, at least for me, it was very positive experience. It was a very good week.

14:34

I felt like they'd done a better job and kind of took some cues from last year's complaints around maybe too many sessions too close together. You know, I found the sessions down at the campground which in the previous year you could hear three loudspeakers going of three different sessions where you were sitting and it was distracting and foolish, you know. So I felt like they're like oh, you know what, let's give them a real space. Let's not try and squeeze you know 50 people into a space for 15 and then have it drowned out by the 100 person session or the 200 person session on either side. So I thought that from an organizational standpoint they'd done a better job overall with that. As far as the mood goes, because I think we can get a sense of a mood for a whole conference, can't we? And I did feel like there was a little bit more I almost I hesitate to say tension, but certainly by Thursday and maybe it's just people are hung over two days in a row and by Thursday they're just like, but there seemed to be sort of some just kind of tiredness and grumpiness going on, at least for the morning, because I talked to a lot of people and I was in a lot of different places and it just sort of felt off. That might be the pink effect.

15:54

I mean, I went to that concert, I stayed for all of one and a half songs and I was like. I looked at my friend. I was like you ready to go? He's like yeah, let's get out of here because pink's not my thing. I knew a guy who was like right down there, front and center. He loved to eat. First of all, he was a fan. And two, when you're immersed in it like that, I think it's a very different experience than the nosebleeds by the concessions, you know, on level two or like whatever. So you know, maybe there's that. I felt like with the Foo Fighters last year, I don't know, the next day everyone's talking about it After Pink. Everyone was just sort of like, yeah, whatever.

16:32 - Fred Cadena (Host)

Which is funny because I think that was like Dave Grohl's fifth time or fourth time playing at a Dreamforce, and this was first for Pink and Imagine Dragons.

16:39 - Josh Matthews (Co-host)

Yeah, yeah, I didn't even see the. I heard them as I was jumping out of my Waymo to go check them out. I mean that was kind of a cool experience too, doing the Waymo in San Francisco, but that's a San Francisco thing, not a Dreamforce thing. As far as the overall messaging, of course that was focused on agent force and that was the talk of the town and you know it's humans with agents driving customer success together. So you know that whole keynote. I did get to go in and sat down for the majority of the keynote and I just have one little complaint, which is the whole idea around them and I get it. I get the messaging humans with agents or humans driving agents or working with agents or whatever to keep the so to sort of like imply like we're not firing anybody. You know what I mean. It's like we need people to do this. But at to your point earlier, eric, I I mean I think, let's face it it's going to reduce the need for certain employees. I don't think it's necessarily going to mean fewer architects or developers or even Salesforce admins.

17:52

I had a nice conversation with Ben McCarthy from Salesforce Ben about AgentForce and a lot of different people throughout the week, a lot of podcasters, golden hoodies, senior architects, business owners to try and get their take on it, and, in general, none of us really felt like it's going to have a massive impact on the ecosystem and hiring in the ecosystem per se. That time will come. I think that's just around the corner. Right? This is the first step, and I think that the first people that are going to be impacted by AgentForce are going to be in marketing and they're going to be in customer service. Right, I mean, to me it's just like a no brainer, now, that's if it works, right, like that's if it works and for it to work. I had a very. I had a nice sit down with a buddy of mine in some swing chairs and I could be misquoting him, but I think it was something along the lines of like yeah, this is great if there's good data and it might have been you, fred, it might have been someone else If there's good data, what company has good data? Right? And so until that's all cleaned up, this stuff is always going to be a little bit funky.

19:07

Plus, we look at data cloud, which was the big push at DF23. And then I'm looking around trying to kind of catch a whiff of it and I'm barely seeing it. I mean, it's out there, you hear people talking about it. They're kind of curious about it, interested in it. It's out there, you hear people talking about it, they're kind of curious about it, interested in it. But even the sizable data cloud projects that I'm aware of, I know of one that's about an 800K implementation, and some others, even those we're still hearing from the AEs about how it works, great.

19:40

But then when you're implementing it, it's like, oh no, that doesn't work. Oh, that doesn't work yet. Oh no, that doesn't work yet. Or yeah, you can build it, but it's not straightforward, it's kind of weird. So we've got to do all these workarounds. So I'm not a huge fan of early product releases that are fully untested or even keynotes where it's clearly been a recorded. You know it's clearly been like a, a recorded example I have. I have flashbacks to Gemini, right, and I and I was catching a little bit of a whiff of that there. So I'm not a huge fan of that stuff. Um, we certainly got it when we looked at the rabbit, didn't we?

20:18 - Fred Cadena (Host)

It was. The rabbit experience in hand was very different than the uh very different than the pseudo setup on stage performance.

20:27 - Josh Matthews (Co-host)

So I'm always a little bit okay. You know grain of salt, let's see. We'll see jury's out. It's gonna work, it's gonna be implemented. It's gonna work, yeah, but how well it works and how fast it works and how many people have to basically get burned on the way towards that path, I don't know.

20:44 - Eric Cook (Co-host)

Yeah, one thing that jumped into my head. That was just kind of a cool feature that I'd never seen done at this conference. You talked about having a little bit of overlap and hearing other sessions, which I know can be very distracting. At this session they had two stages that were back-to-back in the main exhibit hall, but if you wanted to sit in on one of those, you actually showed up and you got headphones and you were either the green team or the red team and it was kind of like a silent disco where everybody that stood around the only person that was talking was the guy on the stage into a mic, and then everybody essentially had their headphones to be able to listen to that content, but everybody else that was milling around at all the different booths and the vendors were not having to fight with that conversation. And that was the first time I'd been to an event that I'd seen that.

21:32 - Josh Matthews (Co-host)

They definitely did that at DF2. And what was nice is I think you could pick your language, so it's sort of like looking at the UN when they're all talking, listening on their own translator headphones.

21:46 - Eric Cook (Co-host)

Yeah, that's pretty cool.

21:48 - Fred Cadena (Host)

It is pretty cool. Yeah, I think I agree with a lot of those observations, josh. I mean, I think and I'll go one better I didn't attend a single session at Dreamforce Now subsequently, in fact, I didn't even attend the keynote. I was on a plane Tuesday morning when the main keynote was happening. But the great thing about it is, you know, through Salesforce Plus, you can log in and you can get, you know, pretty much all the content that you may have gotten at Dreamforce on replay, pretty much everything other than some of the roadmap sessions and I think some of those campfire sessions and stuff aren't in there but the vast majority of the stuff's in it. And I've, you know, subsequently consumed a lot of content on my trip back and over the weekend. But you know, ai was definitely front and center, one of the things I found, you know, interesting. You know this agent force. You mentioned the human agents. I definitely noticed a lot of the same things you did.

22:49

I spent some time Thursday at your kind of observation. I kind of felt like the last day of Dreamforce has always kind of been a little bit light as far as people and I felt it was even more so this one. I did have a few things on my calendar on Thursday that I wanted to make sure I stuck around for, but I did take some time to go and actually get a hands-on experience with the new agent force and I think what I'll kind of say coming away from it I'd seen some of it beforehand and you know it doesn't feel like an order of magnitude change from what you could do before Dreamforce, and I know there's more stuff coming down the pipe. They're going to continue enhancing it. But it is good. It is interesting To your point.

23:40

You know it's only as good as the data that you have. I'll say I differ from you in how much I feel I saw Data Cloud, and maybe it's just because Data Cloud kind of underpins all of this functionality. If you're not bringing that data into Data Cloud, it makes it very difficult for a lot of these other things to work, and so I think that, rather than it be talked about as like its own separate thing, I think data cloud's kind of gotten to this point, at least in Salesforce's mind, if not in their clients' minds that it's just part of the platform. But there were some cool new additions to Data Cloud that were announced. I don't want to get too technical, but more out-of-the-box connectors, some more format support or some more support for unstructured data formats. The real-time is becoming even more real-time, so there was some cool stuff from there. Even more real time, so there was some cool stuff from there.

24:48

The one thing that I found the most interesting with AgentForce, and the thing that's probably different and maybe it's the first place that I'll kind of ask a question of the both of you and it's going to be an unfair question because you'll have no way of knowing the answer, as I don't either but Salesforce put in their announcement around AgentForce, a new piece of functionality that they're calling their reasoning engine, and they're calling it Atlas, which sounds very similar to what OpenAI is saying about.

25:18

Strawberry is a piece of functionality in their AI that allows you to go through, evaluate, reason and refine, and so that is the secret sauce that Salesforce has put into the AgentForce product to make it able to be a little bit more autonomous. It's not just taking a request and spitting an answer right out. It is kind of going through some level of reasoning. And so my slightly unfair question is do you guys think that's something Salesforce developed or do you think they're leveraging Strawberry, and the two pieces of evidence that I kind of will point to is one Salesforce did come out saying, when they first started going down the AI path, that they were partnering with OpenAI and Anthropic and others and not investing a ton in building their own models. For one and for two, I think Strawberry was announced publicly about a week before Dreamforce came around, and so the timing just seems a little convenient from my perspective.

26:35 - Josh Matthews (Co-host)

Yeah, I mean, it's a fair question. I'd be shocked if Salesforce put together a huge team to create something that's competitive with Strawberry. I mean otherwise, otherwise they wouldn't just hoard it, they'd market it and then they'd, you know, just spin it off and go public with it eventually as a separate entity. Because if it's that powerful and competitive, then why wouldn't you and they were? They were sharing things about, you know, roughly a 30 reduction in ai hallucinations, right, I haven't looked at the numbers for for for strawberry, but is that in line with what they're saying about their new engine?

27:12 - Fred Cadena (Host)

because that's the only bit of information that I'd be like that, that's.

27:15 - Josh Matthews (Co-host)

This is the only fact that they gave us. They gave us one number. You know, yeah, they talked about time and real time and things like that too, but they really only gave us one juicy number that can be compared.

27:30 - Fred Cadena (Host)

I'm not suggesting it's a bad thing. I came out when Salesforce was first this was, I think, last Dreamforce I was very forceful in saying I think it's smart for Salesforce not to try to jump in the also-ran category of let's go build our own large language model. And so I'm not saying it's a bad thing if they did. I'm just think it's an interesting question. Sorry, eric, I know I cut you off.

27:54 - Eric Cook (Co-host)

Nope, I was doing a little multitasking there. While Josh was responding, I went out to our favorite AI driven search engine to ask whether there's any evidence that it could find me in perplexity whether or not GPT-01 Preview is powering anything on AgentForce and at this time there's nothing available that the platform has been able to find. So it thinks that there's separate initiatives. But to your point, josh, I don't remember the statistics but during the discussion because, again, it released literally two hours before we were scheduled to finish Mekon but there had been some inklings as to what Strawberry was going to do and why it was going to be such a big game changer and the whole concept. If you're listening to this and you're not exactly sure what, o1 Preview, formerly known as Strawberry, and then QSTAR, even before that, the concept that we've all been operating under as it relates to these GPTs and large language models is these tools have gone out and scoured the internet, the world and the analogy of a librarian that goes to every library on every continent across the entire world and reads every book and consumes all of that information and houses it within a large language model and has developed a prediction engine based off of the GPT framework generative, pre-trained transformer that gives it the ability to say this is what likely is going to be a response, whether it's a written response or an image response or an audio response or a video response. But at no point did any of this technology take into consideration what was going on behind the scenes, in the mind of the creator that was writing that book or doing the painting or composing the music. They only have access to the end result, the finished product, have access to the end result, the finished product. And so what? Gpt-01 Preview? And I think what this agent force is trying to do is get in the head of the human to try to think about well, why did this book come out the way that it did? And maybe we don't want to get in Stephen King's head, because that might be a really scary place, but what was going on inside of Stephen King's head when he was coming up with this story? How did he approach his storylines, how did he develop his characters? How did he come up with the plot and the twist and the end? Those are the things that, when the model is forced to take its time to be able to reason, to be able to check its own work.

30:27

There's a couple of examples on GPT-01 Preview that you can run it through a very complex mathematical equation and you can click the drop down to see what's going on behind the scenes. And this particular example I don't know if it's still out there, but it takes probably close to 40 seconds Because you can look behind the scenes and it's running calculations, it's verifying the calculation, it's making sure to check other resources and if the calculation is not right, it does it again. You have the ability with prompt engineering to be able to go in. One of my favorite prompts is you know, is there anything else that you need to know before you complete the task that I've asked you to do? But there's also an element where people were starting to say take your time, think about this, you know, pause and reflect on the answer that you're about to give, to determine its accuracy and relevancy to the question that I want. So there were ways that you could put that into your prompt process. That would force the AI to go back and go. Okay, I'm going to think a little bit about this.

31:28

But now GPT-01 Preview has actually got that baked into the process and it takes longer. For most queries, 4.0 is probably still going to be just fine. I still go to Claude for a lot of writing because it still does a much better job at writing in human language. Clawed for a lot of writing because it still does a much better job at writing in human language, but for complex tasks and being able to just look and see what's going on inside the brain. I don't know if AgentForce has that transparency layer where you can actually click a box or drop down and see how it's thinking and debating itself before it spits out an answer, but that is an exercise for those that are listening, that have a paid version, to go to a one preview and use one of the tests and click the dropdown and just see what's going on, cause it's what's happening in our heads when somebody asks us a question. We just don't blurt out an answer, we think about it, we ponder what's relevant to this and creepy, but cool, I think, I think.

32:24 - Fred Cadena (Host)

And uh, it was really uh, an interesting discussion towards the end kind of bring a lot of uh, what the two and a half days were together when, when paul and mike talked about it I I don't know if either one of you had a chance to use much of the the o1 preview yourselves I was struggling to think about something I wanted to really test out with it that I thought would be good and not like using a prepackaged example, and actually it's funny. I've been trying to get a change in CRM here, where I am now from a different, very antiquated system. It doesn't have to be Salesforce there's nothing more modern of which Salesforce could fit the bill and one of the things that I'm trying to do is I'm trying to like put together a what I would call kind of a conference room pilot, you know to to to show people what this could look like, because part of what I think is holding us up is just the people not understanding how differentiating moving from a heavy desktopnet server client server-based CRM to a cloud-based CRM would be, and so to do that, I'm still a dyed-in-the-wool kind of consultant type person, and so I'm like, all right, I'm still a dyed in the wool kind of consultant type person, and so I'm like, all right, I'm going to put together a list of user stories and I've been late. Number one I haven't been hands-on in consulting in quite some time. Number two I had been lazy for quite some time and in all the consulting firms that I worked in, you know, had people build giant libraries of user stories and so you're never really starting a project from scratch. You usually, you know, could go in and open up like okay, here's use cases we want and here's user stories for those use cases, and then you just kind of modify them. And so I was like, well, I don't have access to that anymore because I'm not in any of those consulting firms, so maybe I can get you know this, this for this 01 to to to come up with some. You know something good to start with. And so, like I walked it through this whole long prompt talking about user stories, talk about the use case, and I was, I was pretty impressed.

34:40

It came back with 50 starter user stories in proper format. It figured out what the user story format was. It figured out all the fields that you know, the kind of you know dozen or so fields that you need in addition to just the story. It figured out, hey, we need acceptance criteria, we need a business purpose, we need prioritization so people can go through and rank it, we need a story point so people can figure out how much effort there is Right. So it figured all this out and then it spit it all out. I said give it to me in a, in a CSV format so I can just put in Excel.

35:18

And it it took maybe I don't know 45 seconds, maybe a minute I didn't sit there with an egg timer but definitely way faster than if I wrote them. And they were every bit as good as what I would expect to go to one of these kind of generic user story libraries that I used to have. I mean, they're not perfect, they all needed some work, but it was a good place to start. So I was pretty long story short. My one real use of Strawberry. I've been pretty impressed.

35:49 - Eric Cook (Co-host)

Yeah, I've been diving in and giving it just some strategic thought. Build a plan digital adoption how do you approach an organization that's resistant to AI? And helping to identify and uncover fears, potential, you know rejection reasons or why they wouldn't want it and think about a logical argument of why that's not founded in reality, it's based in ignorance. And how do we combat that with knowledge and what's really going on and so being able to have that conversation and dialogue with it, because I think the power of any large language model isn't a one and done either. You've got the ability to spark a thought in a process and then be able to have a continued discussion around that particular item, and you're just going to get a deeper thought process with the 01.

36:49 - Josh Matthews (Co-host)

Yeah, I'd like to ask you a question, fred, about Dreamforce Just sticking with DF for just a minute here.

36:55 - Fred Cadena (Host)

Yeah, sure.

36:58 - Josh Matthews (Co-host)

What was your least favorite part or moment of the conference?

37:03 - Fred Cadena (Host)

Ooh, so I mean this is going to be? I'll give you a more substantive answer. Second, but honestly, it's the location. I just I don't like. I don't like San Francisco.

37:16

You know, I was, I was very I say, lucky, I obviously I paid for it, but I I stayed, you know, maybe maybe two and a half blocks from Moscone and it is a very like sanitized part of San Francisco during the few weeks of Dreamforce. But just, san Francisco is just not a city set up for that conference, be it. From an overall experience of the city, it was very clear if you went outside of that bubble how dirty the city is, how run down it is. Spent some time there on Friday and went to some of the shops. You know half the shops in the area are shut down. You know things that have been there for five or ten years and the ones that were still there either had guards you know as you're walking into each individual store or they had the doors locked and they're like we're open, just knock on the door and we'll come, let you in.

38:17 - Josh Matthews (Co-host)

We'll let you in the whole flash mob, you know, bust down, bust and grab Stuff that the city has experienced, including, I think, Nordstrom, including.

38:28 - Fred Cadena (Host)

Apple Nordstrom gone now, by the way, Not there anymore.

38:32 - Josh Matthews (Co-host)

Well, yeah, they shut it down. A lot of companies are trying to get out. I happen to love San Francisco. I lived there for a while back in the early 90s. There were a lot of reasons to leave and there are all of the reasons that you just shared to leave and there are all of the reasons that you just shared. It was you know you could have a violent block right next to a nice, perfect, amazing block and you know there was just no, no way to guarantee, very few ways to guarantee, that you're going to live in a really good, safe area. You basically have to be at the top of a hill in a very big house, right, because people don't like to walk hills, so that's what you would have to do. But then, of course, when I got out of the area, traditionally after, I have family in the area aunt and uncle and another aunt and cousin. I love getting to see them.

39:24

And the same thing happens to me every year. I land, I get out of the Oakland airport, I breathe the air, I get out of the Oakland airport, I breathe the air, I get on the BART, I come out and I have this wave of nostalgia of like, oh, california, man California. And then even over the weekend they live up in Nevada, which is the north part of Marins. We went up there and we went out to Sonoma and I'm just like, oh man, california. I remember these days this was great. But it doesn't take long downtown at all to realize you know just why you really shouldn't live in the city. And it's all the reasons I left Portland, are all the reasons I'd never move back to San Francisco. It's the same situation.

40:06 - Fred Cadena (Host)

Yeah, and I love California, like in general, and I have, you know, I spent a lot of time in Napa and Sacramento and San Diego and LA and for years I would have told people LA is my least favorite place, mostly for the gridlock more than anything else. But recently I used to love the city. I spent a lot of time in San Francisco between my time when I was at Schwab and all the years I was in the Salesforce ecosystem. Lots of time in San Francisco. I've just not enjoyed the city. I know that wasn't a substantive answer to your question. No, it's fine.

40:45 - Eric Cook (Co-host)

You didn't say pink.

40:48 - Fred Cadena (Host)

I didn't see pink. I didn't go to the concert.

40:51 - Josh Matthews (Co-host)

What about your favorite part of the concert?

40:53 - Fred Cadena (Host)

Let me go back and give a substantive answer, which is this, which is it is kind of the flip side of an observation you made and I don't have the answer to it, because I don't have the answer to it. But your point about hey, things weren't so packed in there weren't the complaints about so many sessions and so much overlap, and I'll say this I know it's kind of like I didn't go to a lot of sessions. I didn't go to any Dreamforce sessions. I did go. You know, partners have stuff set up in the perimeter and they have their own sessions. I did go to a few of those.

41:31

But part of the reason I didn't go to sessions, other than the fact that a lot of them are available for replay, is there just wasn't a lot of sessions that I was particularly interested in going to. I think, in part of the thinning stuff out, we got left with a lot of sessions that just looked kind of vanilla, Like you know, if I were going, as I am, as a person, hyper-interested in banking, and you wanted to see some really kind of nitty-gritty data cloud use cases or agent force use cases, or you know what are people doing, new and interesting, there just wasn't a lot of that content.

42:12 - Josh Matthews (Co-host)

Unless you're FedEx or Aston Martin, you weren't going to get an example.

42:18 - Fred Cadena (Host)

There was one. I watched it on the replay. There was a really great session. I'll give them some credit. I think it was CIBC was the client. It was a breakout session and they talked about how they were using Salesforce AI in their case resolution. But that was one story, one use case, one example, and I remember pre-COVID, in the days that Dreamforce was almost always above 100,000 people. There was a lot more content and I just wish there wasn't just the richness of content that there had been in years past.

43:00 - Eric Cook (Co-host)

That's probably my but again, is that a factor of attendees?

43:04 - Fred Cadena (Host)

I don't know. I mean, I think Josh's point was correct in that there was a lot of feedback from people like it's just too much. You know, I don't like that. I can't see everything. Things are too spread apart and like I mean, there was a year right before COVID where they had people, they brought a cruise ship in and parked it down near closer to Fisherman's.

43:26 - Josh Matthews (Co-host)

Wharf. That was DF-19.

43:27 - Fred Cadena (Host)

Yes, yeah, and they had people staying there. They had, like you know, they took over all of Moscone, all three buildings and the marquee basement and the Intercontinental, and you know.

43:40

then they had other sessions and half the other. You know they had sessions all the way up in Union Square from there, which is not super far, but I mean you don't want to. If Union Square from there, which is not super far, but I mean you don't want to, if you have one session in the Moscone, you don't want your next one to be at the West.

43:53 - Josh Matthews (Co-host)

Yeah, you're going to skip the session getting there.

43:55 - Fred Cadena (Host)

Yeah. So like I get it, like that was too much, and that's one of the reasons why I say like, in addition to the things I said before that were not substantive, I just don't think San Francisco is the right space to set that up, as opposed to a place that has larger dedicated conference space where you can, you know, have more tracks and and more data and make it easier for people to get through in between different sessions I just what would be an example of one of those cities vegas.

44:23 - Josh Matthews (Co-host)

Hate to say it, vegas would be one I mean, I I think one of the benefits of San Francisco, though, fred, is the ability to have these breakout sessions. You can rent out this club, rent out that club, rent out that. Cercante's got a bar and restaurant for two whole days. Salesforce Ben has a bar and restaurant for two whole days. Accenture does the same thing. All of these companies you can walk to all of them. You don't need a cab, you, you can just walk there. And I think that places and I don't, I'm not, I'm less familiar with the, with the vegas conference center, much less than both of you because you've been there, but it's the like. Can you, can you do the same thing? Like, are there enough places in a town that you can just walk to and have all of these outside sort of satellite events? Because I think that's one of the special things about dreamforce it would be.

45:22 - Fred Cadena (Host)

It would be much harder to pull off that, that kind of stuff in vegas. There there's some. Yeah, agreed, there there's there. There's not enough to meet the demand and there's not. I mean, it's funny. Like you say, you can walk to all of them. Part of the reason why you can walk to all of them is so many businesses around there have gone out of business and there's empty spaces that didn't exist five or six years ago. Because five or six years ago, when I was with Silverline, it was so hard to find a space. We were halfway to the Transamerica Tower. It was the closest space that we could find and we weren't necessarily you know, we weren't writing a small check, but it was a quarter of a million people compared to 50,000. It was.

46:04 - Josh Matthews (Co-host)

It was.

46:06 - Fred Cadena (Host)

It was a quarter million people and the real estate was much harder to get into in the boom and bust cycle in San Francisco. No, I mean, I get the point. A lot of the things that make Dreamforce special would change and that's why I'm not necessarily saying change it, I'm just saying it was. If you're asking me what I was most disappointed in, it was the. If I put myself in the shoes of like I'm super in the weeds in my industry and I want to see how all these things are going to apply to me.

46:38 - Josh Matthews (Co-host)

Yeah.

46:39 - Fred Cadena (Host)

I think I was going to come away wanting more than I would have five years ago.

46:44 - Josh Matthews (Co-host)

Yeah, I completely, I completely get that and I felt that way a little bit about career development outside of technology. There were sessions about that. Some of them you know, I know some of the people and I thought Vanessa did a terrific job on 10 questions to ask the stakeholder before you, you know, like add a new, add a new thing to the menu, you know whatever like to the pick list or whatever it was, and it was great you're showing your technical chops there, josh.

47:18

I know, dude, I yeah, I'm just a dumb ass like ask me about behaviors, let's talk about that, and then I won't stutter so much. There was some good stuff, but some of the people that were like chosen to give these sessions, I didn't see them so I can't really comment. But it's like really Like that's who you got, you know that's who you got and I just know it's like to your point, fred, a lot of vanilla, just a lot of the same old shit. Like I could have Googled this, gpt could have written this presentation for me, kind of stuff. And I like stuff that, as you know, that's a little bit more truthful, honest, transparent, down to earth and is probably going to offend some people and boohoo for them, but that's really what I prefer. I think it's very hard to find it at a corporate event, at any corporate event, right, much less one of the biggest tech conferences, but I'd expect to see a little bit more than than I do just doing a sniff test.

48:12 - Eric Cook (Co-host)

it was all a little bit lemony, you know yeah no, I, I would say to to counter that, not that I want to say my conference was better than yours, but it was but your conference was better than ours, is that well?

48:25

it was only: 1100

49:09

So it was a lot of really good info. Even the tech demos they were short, little 15 minute, but even the tech demos that were in the exhibit hall with the earphones were really good and really valuable. They were obviously talking about a product, but all of them focused on something that their product added as a value element, instead of hey, look at all my shit and buy it and here's how much it costs. It was like here's a legitimate problem that we have solved and how AI is playing a part in the efficiency and the ability for it to do its job way faster than ever before way faster than ever before.

49:51 - Josh Matthews (Co-host)

Well, I would think, in general, people going to a Marcon event would be better skilled at conveying their message. Yeah Right, like I mean by nature, and the only other thing I'll say about that is Eric, you're you're a very unique guy. Like I don't know that I can point to many other people I know who just have a thirst to consume new information about what's going on.

50:14

Right. I mean like, just let's kind of admit that, like Fred and I, I feel like Fred and I and if I'm wrong, fred totally correct me but I feel like we sit back a little bit more and say come on, impress me. Do you know what I mean? Versus, like I'm going to check this shit out and I'm stoked about it. Like let's see what's there. You're going to uncover way more wow than we do, cause I or I'll speak for myself cause I'm just sitting on my front porch waiting for the fireworks on the other side of the Hill to show up.

50:43

I'm not driving to the. I'm not driving to the field. You know what I mean. So things are going to be more impressive to you up close and very involved, and I think that's a wonderful if no one's ever recognized that on this show before.

50:56 - Fred Cadena (Host)

No, I appreciate that.

50:58 - Josh Matthews (Co-host)

You got to recognize that this guy is very unique and very smart. How do I send a?

51:03 - Eric Cook (Co-host)

virtual hug.

51:05 - Josh Matthews (Co-host)

You know what let's do this? Let's just both put our hands on the glass like we're in prison there we go, there we go. For the audience. We can see each other, even if you can't see us. Nice, nice.

51:16 - Eric Cook (Co-host)

All right. No, that was really nice. I appreciate that, oh Michael.

51:24 - Fred Cadena (Host)

So you asked another question, Josh, which was what was my Hit that express for those who didn't know what that was what was my favorite part of Dreamforce and I'll say this, and I'll start with a bit of a mea culpa which is I did not approach Dreamforce this year with my typical level of preparedness for a number of reasons, even though you gave me your guide. I just had so much going on with clients and other stuff and I was even at a client event on the other side of the country Monday, so I wasn't even there for half of it on the first day and I did have some stuff planned going in. I wasn't going in a completely blank page, but just the serendipity of that many people in that close, confined space. I'm going to go with six or eight unplanned, unscheduled, unforeseen client conversations, clients that I knew were probably going to be there but I hadn't reached out in advance. I didn't say, hey, let's schedule a lunch or a dinner or a drink or what have you, and just ran into them and had meaningful interaction.

52:43

We talked about the content or we talked about stuff that we're doing together. We talked about whatever. In some cases it was over a drink or over coffee and in other cases it was more just, kind of in passing. But but that was great. You know, I worked in the ecosystem. You know I mean between between a client, you know being a Salesforce customer myself and then being in consulting. You know it's been 16, 17 years of being in the ecosystem, so I know a lot of people. You know, just running into people, that I've worked with, people that are you know off at different consulting firms now, or people that went from the client side to consulting or consulting to the client side.

53:20 - Josh Matthews (Co-host)

It happened when we were hanging out. Absolutely yeah, At least quite one of them scared the crap out of me. He shouted your name so loudly. I was like ah.

53:30 - Eric Cook (Co-host)

Fred's a big deal, man you see him and you're like, here we go.

53:35 - Fred Cadena (Host)

I don't know about all that, but no, it was just great. Like I mean even not that I recommend anybody going into Dreamforce without a plan because you know it's not a cheap ticket and the hotels aren't cheap and the flights aren't cheap. Nothing's cheap about it. Like if you're going to Dreamforce you know whether it's you paying the freight or your company paying the freight. It is hard to get in and out for under four grand by the time you buy a ticket, get a hotel, get around town, dinners and all the rest.

54:04 - Josh Matthews (Co-host)

Yeah it's the dinners that the extra grand is the dinners, right? Oh yeah, but yeah, I was figuring if you're just going but you're not whining and dining, it's three grand.

54:16 - Fred Cadena (Host)

Yeah, but whether it's three or four, right, if you're going to try to cover that kind of spend, go in with a plan. Don't not go in with a plan, but just also know that there's so much there that you're going to be able to make stuff happen no matter what, and and that's. That's the thing, that that you know, part of the magic that's never left dreamforce is is. You know, then, that that would have to be my favorite part of the conference.

54:41 - Josh Matthews (Co-host)

I think that's awesome and and sort of. It's sort of like imagine, imagine you're that all of the Disney parks are in one area and you have a four-hour pass, right, like a three-day pass. You're going to get a lot done, except for the lines. So imagine you've got a fast pass, but for like four hours. That's probably what it's like going to Dreamforce for three days. Yeah, you know, there's just so much going on.

55:09 - Eric Cook (Co-host)

Well, and I was going to say that the favorite part that Fred just conveyed goes back to the whole the uniquely human, non-synthetic moments that are serendipity, when you're walking down the hallway and you're like, oh my God, I haven't seen you for you know years. What have you been up to? What have you been thinking. That's why I think these types of conferences need to continue to live on. They might get smaller, they might change venues, they might go someplace different, but I think if an organizer misses the fact that those little gem moments aren't worth it to them anymore, I think that's going to be a really significant, poorly articulated assumption on their part. It's going to miss out. Significant, poorly articulated assumption on their part. It's going to miss out. So I hope they continue because I love them.

55:55 - Fred Cadena (Host)

Yeah, I'm 100% aligned with that. I am going to bring it back a little bit to some of the AI topic and I'll reference an example I gave a second ago and tie it back to one of your observations, eric, which was still AI with the human in the middle, a theme at Macon and I'll also talk about the AI without a human in the middle in a minute. But that one session I mentioned that had the case resolution workflow. It was definitely built with that in mind. The whole idea was you would have, you know, your customers come in, you know, with a case issue, you know, and through some kind of call deflection strategy, whether it's you know, a form on the website or they have a place to submit a case, or maybe through a chatbot or what have you it goes into a case record and it's not solved at that moment of taking the case in. It's something that requires some level of ongoing work and review, research. What have you? But then when it finally gets picked up I shouldn't say finally when it gets picked up by whoever it is a tier one, tier two support or a manager, or whoever it gets routed to in addition to what the customer has put in the case it's already gone through and leveraged knowledge base, leveraged other case resolutions leveraged, what have you and put together a candidate solution Now it's not so good yet, or they don't recommend now that you just turn it on and have it, push that solution out into the world. But at that point whoever's working to resolve the case is reading that candidate resolution and basically deciding yes, this is good, and maybe tweaking a little bit around the edges or what have you, and hitting a button and sending it off, or they're saying, no, this isn't the way I want to go. And then they're changing it around and it reminds me very much and obviously a lot more sophistication, because the process that I had at Options Express was not understanding and processing real language and making these kinds of queries.

58:25

But in the brokerage space we've got to review accounts before they can trade. You've got to make sure you're assigning the right trade level, commiserate with the risk that the person is say they want to undertake in their financials, and what have you? And the old way or the classic way of doing that is to have somebody sit and look at every application and we decided fairly early on that that was pretty inefficient and we decided fairly early on that that was pretty inefficient and you could take some number of application attributes and put 50 of them on a page and have the computer assign a trade level, a candidate trade level, and then have somebody go through with their own eyes and just validate yes, yes, no, no, no, yes, yes, and you know, kind of go through it. And it just reminded me very much of that same kind of process, just updated with a lot of the kind of LLM and language processing flair. And I'm curious. I'll just throw this out there. Number one, like what do you guys think about that?

59:27 - Eric Cook (Co-host)

And Eric, for you, since you were at Macon, is that a lot of the type of stuff they were talking?

59:36 - Fred Cadena (Host)

about, or were they thinking about it differently? There was a lot to unpack there, so I'm trying to talk.

59:39 - Eric Cook (Co-host)

Yeah, I talk a lot. Can you give me a concise question of what it is that you want me to?

59:42 - Josh Matthews (Co-host)

respond to. For once, I'm the one asking that question.

59:45 - Eric Cook (Co-host)

Usually I'm the guy that rambles on and I'm like what the hell did you just say, Eric?

59:49 - Josh Matthews (Co-host)

e an essay test question from: 1988

59:58 - Eric Cook (Co-host)

Hey, Josh, reboot Fred's avatar. I think he's gone off the rails. Would you control alt? Delete him for me.

:: 00:05

I can do that. Three pencils later, 10 pages later, I hope I get a C on this.

:: 00:12

I guess here's the thing. How were the presenters at Macon thinking about humans in the middle?

:: 00:20

Yes, okay. Yeah, I thought that was where you were going, but I wanted to make sure. Why didn't you just say that?

:: 00:26

man.

:: 00:26

It sounded so good.

:: 00:30

You should have stuck that in the Hemingway app. You could be like just voice to text the whole thing and then click the button Hemingway, this and boom yeah exactly.

:: 00:41

They each had their different spin on it. Language model session that I sat in on when they were talking about humans in the loop and kind of in the loop at the beginning, because it's humans that were populating the language model with the content that wanted to actually be used as part of the process, and then humans in the loop at the back end when the actual result was created, to determine whether there was hallucinations, if it was on brand, if it supported the personality of the organization. So I think you're going to see that human in the loop on both ends of this and the notebook LM thinking about human in the loop. On that, I created a notebook and loaded up the 50 resources, one of which was my probably 25 page note document with screenshots and links and bullet points and other stuff. But I downloaded all the PowerPoints and as many of the session summaries as I could. Now, notebook notebook LM gives you access to 50 documents up to 500,000 words per document, so I probably could combine all of those transcripts, summaries, into one document and call it good.

:: 02:00

But human in the loop on that was me finding all the information. The notes that I took were my notes. Granted, I was using my Plod to be able to also listen, and it generated summaries for me that I could go in and take a look at. But what was created there in my Google document was my insight and observations and takeaways, and then I put that into Notebook LM, which that doesn't use any sort of you've got to populate it with stuff or it just can't tell you anything. So you are creating the language model in this sense and ask it questions.

:: 02:39

I had to generate a 12-minute podcast summary that I listened in the gym this morning to, just to say what did you get out of all the information that was at Mekon? And it missed a few things, but it was 12 minutes of I don't know how many thousands of words and hundreds of pages. But they were all very adamant that you need to be there, you need to listen to it, you need to watch it, you need to read it and you need to check it. And you're going to find more platforms, like what perplexity has done, offering up citations and references just to make sure that hallucinating bullshit doesn't sneak into the end result, and you can go back and confirm it, but then you still have to make it your own.

:: 03:24

It's got to sound like eric or fred or josh at the end of the day when I hear, when I hear you talk about it this way but both you guys it it really strikes me you have to have a relationship with it. Yeah, I mean, you really do. It's not just like. This isn't like the relationship you have with your Roomba, you know, or with your, you know, with Fred's, one of Fred's incredible coffee maker, espresso makers. It's not like that kind of relationship. You really are going to learn about it, what it likes, what it doesn't like, what it needs from you, and then you can give it what it needs. It's like the five love languages of AI. Let's write that fucking book, no doubt. Five love languages of ai, yeah, let's write that book, no doubt. Yeah, like, because it's. And I wonder, and you guys tell me what you think.

:: 04:16

I, I'm trying, I've been trying to think like, okay, well, what would that be like? Like when in our lives do we have to learn a new language to communicate with someone? Right? We learn it when we have our experiences with our teachers, right, that's something. We didn't have a teacher before and now we have a teacher, right? I would say your relationship really changes with your professors in college, depending on the school you go to, but definitely it changes. You're an adult. This is a different relationship. We're your first employer of a serious place. I don't mean Debbie from Subway who's going to yell at you all day, and she really shouldn't be a manager. I'm talking about directors, vps, presidents.

:: 05:00

Sorry to all of our sandwich artists who are listening to this podcast. No offense, yeah, hey, keep it up. Keep it up. Gluten-free tuna Six inch let's go.

:: 05:06

Keep it up. Keep it up. Gluten-free tuna Six inch, let's go. So you know like. But I wonder, is it like having an employee for the first time and having to learn, Because we all know that takes it doesn't happen in three months? Yeah Right, we all have gone through it. You're a manager for the first time. How long until you got good at that? Do you think it's going to take us that long to get good at our interacting with AI, or is AI going to improve so fast that it's going to shorten the curve significantly.

:: 05:41

Yeah, I would say the, and I talked about this when I came back from social media marketing world because I heard Drew Davis present his digital doppelganger presentation at that event and it was slightly different at this one. He is not to make it sound creepy, but he's gotten very intimate with this avatar that he's created. It has his likeness, it has all of his speech, it understands his mannerisms and how he talks and he shared how, when he gets a speaking request, when he needs to produce a video, when he's got an email, drew Dini, which is his alter ego, actually has his own email address. So you can email Drew Dini at Andrew's company and Drew Dini is the agent. The agent will read it will come back and it'll answer to you and it'll put a footnote. You know will read, it will come back and it'll answer to you and it'll put a footnote.

:: 06:31

You know this message was generated by Drudini, my AI avatar. This was reviewed by Andrew, so you know he takes a look at it before it actually goes out, but sometimes he doesn't, sometimes it's just an automated boom. He's gotten to the point now where his relationship with his second ego, second, second entity, has gotten to the point where he trusts it now to be able to do things wow we were just talking about this on the last show.

:: 06:57

We were talking about when is ai going to slow down our email? And what have you said? Oh, it's one that's responsible for speeding it up. I, I was.

:: 07:06

I was sitting next to our chief AI officer, robert Mitchell, with WSI Corporate. He's going to be here at the conference and we're on a couple of panel discussions here at a WSI event. But he was very adamant in taking notes and he's like I'm going to create a Robert entity. I'm going to create a separate email account. I'm going to configure it with a GPT, be able to tie it in with a make workflow process that taps into OpenAI, with some documents that are examples of emails that I've written and the logic that I want to have behind it and when I'm okay with it sending and when I want it to trigger me for a review, because there's a lot of stuff that we all get support tickets and just common questions that arrive that if you can take that off your plate and provide something that's valuable that's on point then why not? And Drew's a great example of how that's happened. If you get a chance to hear him present his digital doppelganger, he's a masterful presenter. The lessons are awesome. You'll really enjoy it.

:: 08:12

I definitely want to. I was disappointed. I missed it at Social Media Marketing World and now doubly so that I missed it for a second time. But it's interesting and I think, obviously I think it's going to come to this eventually in a more mainstream way.

:: 08:31

And.

:: 08:31

I think I was the one that said it's speeding up your email. I just foresee this time in the not-too-distant future where there's AI SDRs and AI AEs writing emails to AI customers that are just kind of going through their own cadences and nobody's actually talking to anybody else. I don't know if that's a good thing or not.

:: 09:01

Will the checks still clear? That's what I want to know.

:: 09:03

Well, I wonder if the checks will even come right.

:: 09:07

If you basically say, hey, here's my list of 10 preferred vendors. If you get SDR outreach from anybody not on this list, you know, just kind of string them along, be nice. You know I'm a Midwesterner. Now, right, I've lived in the Midwest for 18 years now. So you know, be Midwest nice, string them along, but them along, but let's not get anywhere with them. And how much effort is going to be spun on sales cycles that never should have been? How many people, speaking as a sales guy here, how many people are going to miss out on hearing what could have been a better solution? The only reason I'm in sales is I do think it actually performs a valid function in solution discovery and it's interesting. I mean, it was certainly not a Luddite, and I'm not saying that we should stop it, I'm just saying we're all going to have to take a step back and think of how we're going to respond to these changes coming so quickly.

:: 10:16

Yeah, it's a really neat point you make, I think, fred, and it got me thinking about the old movie Multiplicity and it's Michael Keaton and he finds a way to make a copy of himself so he can get his work done and have a home life, and then the copy makes a copy of itself and you get a copy of a copy of a copy and for those who used to use copy machines, we all know it's just image degradation every cycle and then hilarity ensues. But I imagine the same thing can happen if your AI is just off a little bit. And I was starting to think about okay, eric, you hire an assistant and you train this person to do your sales calls for you. Train them like crazy and off they go. They're going to close half the deals you would have closed yep it just is.

:: 11:13

It's the multiplicity problem and I, I just I can't fathom ai ever being better than you. But could it get it might be percent.

:: 11:24

You know, yeah, it might be 50 well, I don't know, that's what I don't know.

:: 11:28

well, here's the thing, though it might catch things that you wouldn't have, but my guess is it's going to miss out potentially too.

:: 11:36

But here's the question though Is 50% good enough If AI doesn't eat, it doesn't sleep, it doesn't take a week and go to Cancun?

:: 11:45

Hey, now hey, now I'm just saying I'm still here, come on.

:: 11:49

You're still here. How many more at bats are you going to get? Like, if you really turn this over to ai, are you going to get 10 times as many at bats? 20 it's not infinite, right? There's only that many opportunities on your market size. Yeah, right it depends on.

:: 12:05

It depends on your market size, right, because, right, if you, if you have a small market I have, I believe I have a relatively small market there aren't a lot of firms out there that just do salesforce and that's it. Yeah, right, very few. And the reason there's very few is because it's a very small market. The big companies look at it as not sizable enough for them to invest in in a deep way. So there's that. And then the other thing is you have what's the replication within your industry going to be like? So if you now have to send out four times as many messages or have four times as many meetings to get the same revenue as if you're doing it on your own, well then your competitors are too. Now you're turning the whole system into something that, within two or three months, people start to recognize it.

:: 12:59

I deleted tonight, just tonight, I deleted at least 12 or 13 emails from different people that said Josh comma, quick question. Okay, do you see what I mean? That that's tonight, 12 or 13 of them, and those weren't my only. There were other ones that had another. You know catchy phrase, but they're all using it. So what you're?

:: 13:22

saying I should say Josh, really long, complicated question, that would be completely off. I'm joking.

:: 13:32

Anyway just something to think about. I mean, the preseason is worth investing to prove the model For everybody that's an end recipient of that.

:: 13:44

They're going to get four messages this week and then three months from now they're going to get 40 messages, because the amplification, yeah that's. And I guess that's where I hate to keep going back to the human, the human, the human. But you know, is there a way that you can take that in and figure out? Maybe an email is not the right thing, but maybe it can help you write a script for a phone call, or it can help you come up with something else that's creative, that's going to get somebody, you know, their attention, I mean. But here's the thing.

:: 14:19

Theoretically right. If AI gets good enough, and good enough in your voice, it should be able to, you know, do the research on your prospect to get to a level of personalization Again, maybe not as good as I could do, but 50%, 60%, 80%, whatever that gets to, it should be able to rewrite the email. So it's not quick question 50 times, right, it's, you know it's its own language. 50 times, I mean. I think that you again, as these things get better, I think it's going to be more and more difficult for the average bear to discern whether or not it's AI Right, and and and, again, I think. I think that I don't, we can't. It's. It's meaningless to debate whether or not that's a good or bad thing. It is the reality of where we're going. The question is, how are we going to react to it? And then to get back to the human in the middle piece? One thing I kind of decided on this podcast as we were talking about it on this podcast, as we were talking about it, you know, if we're ever going to get to the human, not in the middle, it's going to be something like Strawberry or something like Atlas that is doing the reasoning and that, for the foreseeable future will let you peek under the hood and get you comfortable with how it's getting from point A to point B. You know, the more I think about what Strawberry did for me the other day with that user story example.

:: 15:59

What it really did is, instead of me having to go because what I would have done two weeks ago is I would have gone to perplexity and I would have kind of you know led the witness I would have said all right, here's prompt one. All right, okay, now you know this. All right, here's prompt one. All right, go, okay, now you know this. All right, here's prompt two, here's prompt three. And I could have probably gotten there in eight or 10 prompts.

:: 16:20

To where I ended up, I got there in one prompt. I was pretty happy with what I got. I mean, it was a long prompt. They didn't just write two sentences. And here we go. It was a well thought out you know, if I don't say so myself prompt and it was long. But if I do that 50 times, if I do that 100 times, if I do that 1,000 times over the next several months or year, am I going to get to a point, maybe, where I'm comfortable saying, okay, I know how this thing thinks I'm willing to let go a little bit and put some stuff out there.

:: 16:53

The answer is maybe yes, yeah, yeah, that's freaking dope, dope, dope.

:: 16:59

I want to see this prompt that he wrote, this freaking masterpiece of a prompt.

:: 17:04

I don't know that it was a masterpiece.

:: 17:06

I want to see it too. I want to see this prompt. Show me your prompt. Let me ask you guys something what if this is just uh, this is more sci-fi level ai discussion? Okay, what if in a world, in a world, just kidding in a world? So what if companies signed up to some sort of membership where they really like it's got to be super secure, super safe, but basically you're running your AI on it. But, more importantly, as you're running your AI through this central system, it's learning specifically about your business and what your needs are, and if it does that with enough companies, it can create zero sales. Just 100% recommendations. Bing, eric, your company would increase your profits by 8% or more on average with an investment of $12,000 for this type of consulting for your practice.

:: 18:06

We give this a 98% score. And then it's just like now you're not even. It's just like the ai knows what you need, knows exactly who it should connect you to right and you know, off off, everyone goes. It's like if upwork had no advertising. It just you logged in and it was like yeah, I know, I know, josh, you need a email mark marketing person to set up marketing cloud for blah, blah, blah, whatever. It's like, yeah, how much Bing bing.

:: 18:42

Well, and maybe the question that we've been asking is the wrong question how much better would an AI agent be at outbound calling or generating sales or doing our job? Maybe the real question. To go back to what you just sparked a thought on, josh's, maybe we use the ai internally to help us identify those opportunities and the relevancy so that, in a finite world I'd added finite in a finite world- in a finite world, but in a finite world with limited time and limited resources, maybe AI needs to be more of a behind the scenes.

:: 19:19

I'm going to analyze your data, I'm going to look at the market.

:: 19:22

I'm going to take a look at people and see what their interest is.

:: 19:25

I'm going to identify prospects for you and then I'm going to give you a hit list and then I'm going to get the hell out of the way and let Josh work his magic on the six people out of the 60 that you would have previously been reaching out to, as opposed to going. Can we take 60 and go to 240? And hopefully we get a good one? Now it's like, instead of 60, we've got six and we're prepared as hell because we know everything about them. We know what their likely resistance points are going to be, we know what their objections are likely going to come across, so we're prepared to answer those and we show up better, prepared and intelligent and, at the end of the day, it's the human that is that. That requires a crap ton of work and it's not as easy as pressing a button and saying blast 240 messages out. That kind of sound like me and I'm hoping that I can at least get a 20% hit rate and stay on point. But that's the flip side to that coin.

:: 20:26

So yeah, thanks for sparking that thought, man, it's smarter harder, right, yeah, that's all it is. Totally, totally, totally Smarter always takes more prep, a lot more. Oh yeah, absolutely Not a little bit. A lot more, yeah, like more more prep than people are comfortable giving it to do it well, yeah, yeah.

:: 20:46

And it's not a silver bullet.

:: 20:47

No sure, yeah, it's not cool man, what a neat convo today.

:: 20:52

Yeah, I don't know if we should stop it right here, but I feel like this is a really good drop the mic moment. It is a really good drop the mic moment. I was actually going to suggest we wrap up.

:: 21:03

So conference season is a little bit behind us, but I will ask the usual question. I believe Josh, you're going to be at Florida, dreamin' right I?

:: 21:10

I believe Josh you're going to be at Florida Dreamin'. Right, I'm going to be at Florida Dreamin'. That's on October. What is it next week? Yeah, it's, I think, october 3rd through the 5th. Yeah, 2nd through the 4th or something like that. So, yeah, I'll be there for a few days. That's in Clearwater Beach. I'll be doing a session on Friday morning. It's honestly no preparation, which is great. It's just going to be a live recording of the Salesforce Career Show podcast and be in AMA format. So I'm just going to field questions and hopefully help people get some of their juiciest questions answered.

:: 21:50

Nice Juicy questions Ericicy questions Nice.

:: 21:55

Eric, how about yourself, you?

:: 21:56

going anywhere. Well, when I get back from Mexico and recover from Cancun, I'm off to Madison, wisconsin, for the Graduate School of Banking's Strategic Marketing School, doing a session Thursday afternoon, going to dive in on all things digital, social and, of course, ai. I'm going to be with my banker friends. Last year I introduced a number of them to Pi, the emotionally intelligent assistant, and we had some really fun conversations with Pi up in the study pub a quote, unquote of the Flunos Center. So looking forward to that and then Eric man, you kill me, dude.

:: 22:36

I know it's going to it's going to be awesome.

:: 22:40

And then, not until November, do I really go to another event, fedfiz the FinTech Cowboys. They've got a a FinTech roundup that they have out in Texas and they've asked me to be present at that and chat about AI and field the AI table for bankers, and I am super excited to actually meet Dave and Tanner Mayo in person. They've been guests on our Linked Banker Mastermind a couple of times and I've been on their FinTech Cowboy podcast, so they're awesome guys really, really care a lot about community banking and innovation and, um, just uh, good old texans probably going to get me some barbecue and some cold beer and talk ai and fintech. So that's going to be a fun few days in texas, so that's coming up.

:: 23:27

Are you going to be there during the f1?

:: 23:30

Austin Can't really not sure on that, I don't know. On the F1 event, I'm probably going to be talked into riding a horse, so I'll be on the throttle of a one-horsepower machine there you go, there you go.

:: 23:44

Sounds fun.

:: 23:46

Yeah, that should be a cool event. I know they're like very protective. It's not an event where they let a lot of vendors come in. It tends to be very banker focused.

:: 23:58

You've got like 200 seats and that's about it. So I was very humbled and thankful that Dave reached out, but I really have enjoyed getting to know those boys and can't wait to get out there and have some conversation and meet all sorts of new bankers and rub elbows with some of the cool innovative fintechers that are doing some cool stuff in the banking industry, so I'll be excited. Very cool, yeah, I'm looking forward to hearing your experience out there.

:: 24:26

I'll be in Denver the 13th through the 16th for the Go West Max Credit Union League Conference. It's all the many of the credit union leagues in the West and then I'm speaking there and then the following week I'm not speaking but I will be at the REACH Conference, which is the California and Nevada Credit Unionvada credit union league conference, and that's in uh san diego, so a couple of uh san diego credit union on my, on my horizon.

:: 25:03

Very cool, so cool. Gentlemen, this is a great episode really. This is awesome. I love hanging out with you guys good stuff and uh appreciate the ideas.

:: 25:13

You give me some cool thoughts to think about as well, so I appreciate that. Sweet We'll talk again soon.

:: 25:18

Cool. Bye-bye, peace. Well, everyone, we hope you enjoyed episode 34 of Banking on Disruption. Don't forget you can find show notes and a full transcript of the show on our website, bankingondisruptioncom. This week marked another milestone for Banking on Disruption.

:: 25:38

Back in July, I decided to launch a daily banking news headlines podcast, ss, an offshoot of this bi-weekly banking and technology interview and discussion show. I wasn't sure what to expect. I intentionally didn't promote the daily podcast much, partially because I wanted to see what type of. I intentionally didn't promote the daily podcast much partially because I wanted to see what type of organic traffic and traction we could develop and, in all honesty, I wanted to make sure I could pull off producing a daily podcast. I'm excited to announce that earlier this week, on Tuesday, we released episode 50 of the Banking on Disruption Daily podcast. If you haven't checked it out yet, point your favorite podcast app at Banking on Disruption Daily. And don't forget, new episodes of this podcast drop every other Thursday. So we'll see you in two weeks and in the meantime, don't forget to follow us on LinkedIn and Instagram at at Banking on Disruption. Until next time, this is Fred Cadena, wishing you success in your digital pursuits.