BUILD

SUCCEED

Dan: [00:00:00] you really want to have high confidence and conviction about the problem that you're working on. And I feel the same ethos applies here. Inside of your business, hey, we know there is inefficiency and problem areas in our sales teams and how they operate.

let's go focus on that problem and iterate your way into it.

David: Hi, I'm David, and this is Build to Succeed from Very Good Ventures. Today, we talk with Dan Ho, founder of A. I. Consultancy Eskridge. In this episode, Dan's going to tell us some very sound advice about how to get started with A.

I., both for internal efficiency and productivity, and to improve your digital products. So, without further ado, Let's get into it.

Alright. Hi Dan. Thanks for joining the Build to Succeed podcast. 

Dan: Very excited to be here. I was saying this is, I don't think I've done a podcast before. I think it's my first podcast 

David: Well, Let's uh, . 

well why don't we start, why don't you just give a quick introduction. Who are you and what are you doing? 

Dan: Yeah. My name is Dan. I live in DC. I am the co founder and partner at Eskridge, which is an A. I. consultancy that works with, mid [00:01:00] market companies to deploy A. I. and yeah, that's what I'm up to these days.

David: So A. I., obviously a lot of people are talking about it, so uh, first question, gotta get it out of the way. Do you think we're building Skynet and will the Terminators take over the world? Or what does this mean for humanity? 

Dan: I feel like we're, we gotta be past that question, right? Um, 

David: Meaning it's already like, it's already going to happen. We already know that. 

Dan: I don't think we're building Skynet. I don't think the, you know, the sort of apocalyptic nuclear war happening is, anywhere close to reality. I do think there is a significant possibility that over time as A. I. continues to evolve, it will dramatically impact the work that people do.

In some way, shape, or form, I, I don't think it'll displace people, but it'll change what they do, absolutely.

David: Well, so let's go back to your background. you know, obviously you're deeply experienced in technology and digital products given your, your career trajectory. So I think that's helpful context to how you got here. So I was wondering if you could give us a brief [00:02:00] overview of your career journey and ultimately what got you to being fully committed to A.

I. at this current moment. 

Dan: Yeah, I was thinking about how to, to cover this topic. I feel like you might find this funny. So, we were in undergrad together in this program called Management and Technology. And I don't know if you remember, David, we had to write an essay as a part of the college application. And in that essay, I swear, that's the reason I got into the program, because I don't think my, Creds were that, that strong.

but I wrote about paradigm shifts and technology and how I really wanted to be a part of these waves. And I feel like, so I spent my whole career in tech, but really, I feel like I've been chasing different paradigm shifts. So I was at Microsoft early on, straight out of school, and I went to Motorola to work on the mobile wave.

You could see it coming. I feel like I got the wave right. I chose the wrong company. At the time, Motorola was kind of the hotness. Uh, in hindsight, there was a new entrance to the market that, that I didn't foresee. And then I was at HUGE for many years, [00:03:00] and I got to be a part of like this tidal wave of user centric design and design thinking that was really exciting.

David: and what is huge for people who maybe don't know it. 

Dan: uh, yeah, so HUGE is a design and development agency, uhm, digital agency. It was one of the, one of the first major players to really embrace user centric design. and capabilities have kind of expanded and evolved over the years, but, similar to like a frog design or like an IDEO, I would say, RGA, in that vein.

So I spent a long time there. most recently was at Amazon. and that's when I first encountered generative A. I. Uh, we were, uh, Using very early versions of stable diffusion to try to automate ad creative. And this was back in the days where you could see, you know, if you try to generate a person, there's like a hand hanging off the side of their face or some weird, you know, nightmarish thing that was kind of close, but clearly in, in the uncanny valley.

but it was, um, you know, kind of a skunkworks project and only about. I don't know, a few hours a week, for me and, [00:04:00] a couple of years ago, I sort of had this realization. I think this is going to be big and I really want to do this full time. So that led to Eskridge.

David: So tell us a little bit about that. What is, uh, what's Eskridge.

up to? 

Dan: Yeah. So Eskridge, uh, primarily works with mid market companies and we work with them to figure out, where and how, uh, A. I. can impact their business. I think a lot of, seasoned business leaders are trying to figure that out from themselves, and I think a lot of people are going to make mistakes along the way.

Right now our focus is heavily on the internal operational efficiency side, primarily because that's where people's comfort levels are right now. more back office applications of A. I. where I have a person in the mix who's operating as a co pilot. Any errors or hallucinations you're not necessarily passing on directly to your customers.

I think over time that will start to shift and as people get more comfortable with, I think you're going to start to see more user facing digital experiences that are very heavily leveraging, uh, large language [00:05:00] models. and I think there's a marketing application too. Okay, you're going to love this.

Just last week, for the first time, I had somebody ping me who was like, Hey, I'm getting a couple business referrals. From chat GPT recommending my company and how do I boost the rankings? And there's a whole like new world of AI optimization that is barely scratched the surface of that is like the new search engine optimization.

I think, 

um, 

David: totally. That happened to us recently. Someone on the marketing team said that, uh, somebody. was asking the chat GPT about Flutter. And, and 

we showed up in it 

And we were like, okay, how do we do more of 

that? 

Dan: Oh yeah, you know, I can't wait to monetize this, I'm sure.

David: Well, and then there will be a whole battle in warp of like the GENAI companies trying to get around, the people who are gaming it, you know, 

Dan: Yeah, yeah, 

David: just like what Google has been doing for the last 20 years. 

Dan: Exactly, exactly. Yeah, that's gonna be super interesting.

David: Well, so obviously there's a lot of change here and it's, um, an extremely, [00:06:00] Topical, like very current thing. You said like back in the day, when stable diffusion was, but, when you say back in the day, That was like. Two or three years 

ago. so, uh, that just shows you how quickly things are moving here. Um, there's kind of two angles I'd love to talk to you today about and tap into your knowledge and experience, given that you're working with companies and spending a lot of time thinking about this, you know, there's two angles and you, you hit on them, uh, there's two angles for A.

I. One is I like to call it like the production advantage side of A. I. Like, how do we use it internally for more internal efficiency, things that give us a competitive advantage in terms of operations as a, you know, and then there's the more. Product experience side of A. I., of like, what is the, what is the digital tool, product, how do we integrate this into a product, and kind of like create value for our customers, like, a better experience, and they're oftentimes, I think, kind of looked at a similar set of tools, but kind of different business cases and expectations, Given you were saying, yeah.

you've been spending a lot of time thinking about the like sort of internal efficiency production side of [00:07:00] things, maybe let's start there.

Um, what are some of the most exciting ways you've been seeing A. I. sort of transform internal process and workflow? 

Dan: I think it depends heavily on the team and, and the business, but there are certain companies that I've seen, and I'm working with some of them where A. I. could potentially completely upend how they work end to end. I'm working with a creative production company right now. And, um, as an example, as you can imagine, creating like banner ads or ad social media content for different sizes and formats like that is, is dramatically shifted and changed, through all of the AI capabilities.

there's lots of companies out there as well that, um, rely on video content, you know, podcasts even expect to be significantly, uh, impacted over time. So there's a few different ones that I'm working with. I won't go too much into the details. There's a research firm that I'm working with doing lots of different forms of user research.

I think that's ripe for disruption with A. I. But then [00:08:00] on the flip side, you look at these companies, for example, I was talking to, like a large fitness chain. And, you know, I don't think it's going to be that heavily impacted. There may be some back office things that may shift, like, you know, which is maybe like 5 percent of the business, but the fundamentals of what they're doing is not necessarily going to change over time.

So I think it's very sector to sector.

David: So if a company were to come in and sort of start to think about how am I going to use A. I. internally to improve our productivity or efficiency, what's the framework by which you start to break that problem down? Because I'm on some levels, kind of every team can be influenced or affected 

by it. How do you start to like, tease it apart and zero in on the high impact areas and how to get started? 

Dan: yeah, I think that's the key is identifying high impact. I just don't think of high impact as well as like, oh, you can define it a lot of ways. So high impact, it has a big implication on either cost or revenue, whether a lot of people are doing it or, or it directly drives revenue, like [00:09:00] sales activities, for example, can have an outsized impact.

But we also look at it at a task level. You're looking for highly concentrated tasks, right? Cause the problem is if I save you five minutes here and a few minutes there, like with, with context switching and all of that, it doesn't actually impact your day to day that much. It has to be a pretty significant swath of work, that has to be obviated in order for, uh, at a business level, you to start to really see the benefits.

So, Yeah, I think the concentration of it, the downstream, you know, financial implications, all of that has to be factored in. 

And then the third piece of it is also just even what A. I. is good for, 

David: Right, yeah. 

Dan: right? It doesn't equally apply to all use cases, so.

David: That's an interesting dynamic, and you and I have talked about that before, is um, the balance of people wanting to know what A. I. can do for them, but then the sort of business question of like, well, what kind of business result do you want A. I. to achieve for you? And this sort of infinite loop that happens there where I think business people [00:10:00] don't maybe understand the capabilities of the tools or how to take action with them.

Uh, and they're looking for ideas, but then similarly, the tools can kind of do a lot of things. So it's like, well, what are you actually trying to achieve? How are you helping companies navigate that process? 

Dan: It reminds me a bit of like, you know, anybody who studied UX is familiar with the paradox of choice. I feel like that's one of the challenges with A. I. is you can do so many things like the, the potentials are so vast. It's hard to really wrap your head around it. I do think there's an education component.

I think you hit on it. Like the only way out of that loop is to educate yourself just a little bit. because even when I work with clients, I can bring in some perspective on what A. I. is good for or not. But somebody as in the client or somebody internally has to have that perspective of what are the challenges and unique problems this business has, you know, and you have to somehow bring the two of those together, in order to get at what you described.

So, the easiest path is you just take some time out of your, I'm sure very busy [00:11:00] schedule and educate yourself a little bit on, um, on what it can do or what it can't do. The challenge that I found though, is, um, You can do that at, at like a point in time, but staying on top of like everything that's changing and like the capabilities are constantly evolving that, that's a bit of a tricky, tricky challenge, which is part of the reason why S Grid exists, to be perfectly honest.

David: Well, and there's a constant pressure too, right? You have management or, you know, the board or leadership saying, Hey, what's our A. I. strategy? what are we doing? What products are we launching? What are we implementing? And it sounds like the strategy that's kind of the core of it is like, well, just start educating yourself and just experimenting and learning a little bit. 

Dan: that has to be the start. And then very quickly, you want to pick, I think, one or two very high value impact areas and just go deep there and keep iterating and keep iterating. Um, You know, you're an entrepreneur. I had a SaaS startup previously. and you know, one of the guidelines that everybody follows is you pick a problem.

You can [00:12:00] iterate within how you solve that problem, but you really want to have high confidence and conviction about the problem that you're working on. And I feel the same ethos applies here. Inside of your business, hey, we know there is inefficiency and problem areas in our sales teams and how they operate.

let's go focus on that problem and iterate your way into it.

David: Do you think a lot of companies have a good mental model right now of like the low hanging fruit, big problems that they can go after and solve? Is it.

all just like chatbots and customer service and things like that? Or have there been any, um, like, are there any sort of like latent Opportunities that you see as big opportunity that that people haven't that aren't like the common thing people are looking into Just curious.

What is the range of 

of high impact areas you're seeing? 

Dan: I think the short answer is yes. I think there's, here's the challenge is everybody wants to know like, Hey, this is like the use case, that everybody's doing, therefore I should do. And I feel like that's the wrong approach because most businesses are unique enough that, okay, we know customer service, for example, is a very [00:13:00] common application of A.

I. but a lot of companies, especially mid market companies, your customer service operation isn't scaled enough where this is your top priority. it shouldn't be your top priority. so, It depends, but there have been some very interesting use cases. Um, I spoke with this one company, and this is a good example of where they weren't quite convinced, to move forward.

they did mystery shopping. I think I might have mentioned them to you before, but, you know, mystery shopping, hey, you know, you're, you're the head of operations for Pizza Hut, and I want everybody who walks into Pizza Hut to be greeted a certain way. I want my, Servers to be dressed a certain way. The area needs to be clean and well lit.

How do you enforce that or even measure that across thousands of different Pizza Hut locations, right? You have mystery shoppers, who are compensated for going in. They wear these little like, like hidden mics and hidden cameras. and, and it's all video footage and you record their experience. And a lot of these companies today, the way they operate, is they have armies of people who are [00:14:00] going down the checklist.

Okay, were they wearing the uniform? Yes? Okay, check that box. And they're manually annotating these videos. and the multimodal models now are getting good enough that, I think 85 95 percent of that could be automated. 

David: Just through the video feed that they're capturing and 

Dan: Yeah, let's do the video feed, um, without anything else. Um, yeah, it's, it's remarkable.

And again, those are the things that like, even if it's only like 85, 90 percent now, it's only going to get better. Um, and you should plan and make that assumption that it's going to continue to improve.

David: So I have a question for you. How hard is it to do these things? Right? Um, how expensive is it to implement this stuff? What are the barriers for a company to actually do these things? 

Dan: Man, I think a lot of the answer is I don't think it's that hard to get started at least I think it's very easy to do something. That's at least passable. It can be very very difficult to do something. That's exceptional But for most companies, I think passable is already a lot of value. I think the biggest challenge is You know, there's a [00:15:00] fear of A.

I. that exists and hesitancy around embracing A. I. Whether that's because people are, you know, there's, you started off with the Skynet question, there's a bit of this like existential risk that people feel associated with A. I. and it's something new and novel. And I think the other challenge is it's, It's on everybody's list of things to do, but from what I've seen, a lot of companies, it tends to be the third or fourth thing on the priority list.

so it's, it's in everybody's mind, but whether it's like the number one burning need that we have to solve in Q1, maybe, maybe not depends again on the industry. Yeah, but, 

David: Do you find though, though, that there is a dis, I mean, I, I feel like there's somewhat of a disconnect still in terms of, it seems like these things and the promise is that it should be easy. Like I should just be able to go into like chat GPT and be like, change my business to be more efficient. And like, it magically does it.

Right. but in reality, I think the, the, Maybe what we're getting at is that the technical implementations and the things that are there are reasonable to go after. It's not like you have to [00:16:00] like, you know, create a whole new cloud architecture or like, you know, like going from digital transformation where you have like a bunch of on prem servers and you're moving to cloud.

Those are like big things. You're like moving servers. You're doing a lot of stuff. There's like the tools exist. There's a lot of good tools and you can kind of hook up and build on things that, are out there. To me, it seems like a lot of the barriers in the hard part, the expensive part is the process, right?

Like is your data. Set up correctly, like even picking the right?

tool. I mean, I've been looking at this, even for VGV, some of the things, and it's Like, well, where do you even start? I mean, there's so many options. There's so many tools. There's so many approaches. It's like, should I be doing, you know, something like,

Dan: Like, if you were like, looking at just cloud storage, right? We all know cloud storage is like, oh, man, should I use GCP or AWS if you're just starting? Does it really matter? Like some people will murder me for saying that, but like, at some point you just need, you know, a server and where it's stored is like, okay, it could it be 5 percent better or 10 percent better if you made the right provider decision, but it's not that material compared to doing something [00:17:00] versus doing nothing.

and I feel the same way about like, whether you're using like Cloud or Anthropic or using Pydantic or LangChain, like it kind of, something is better than nothing rather than obsessing over optimizing like the last 5%,

David: yeah. I, although I do wonder if, um, large scale enterprises are sort of set up for that in terms of their decision making culture. Like my joke with A. I. is that despite all these new things that are coming out all the time, A. I., the A. I. wars will be won by the procurement teams of very large businesses.

Like the same people who pick, we're going to use Microsoft Office and Microsoft Teams, or we're going to use like Google Workspace and Slack. Like whoever makes that decision for 10, 10 employees, the same person's going to be like, we're going to use Microsoft Copilot. We're going to use Chats GPT, right?

and that's just like how business culture and decision making happens. So there's a bit of a. Thing that is at odds there, which is like, just get started, just experiment. But then you have this corporate culture of like, we have to have the right answer. That's going to scale for the next 10 years. 

Dan: and what I'm seeing emerge at a lot of enterprises, and I've had a pilot [00:18:00] that kind of ran into a bit of a stumbling block with like, uh, you know, top five retailer. is there's these like internal A. I. review committees now that are, you know, it's like a committee from across, you know, fortune 50, and they're making decisions about what each individual department can or can't use.

And on the 1 hand, you need governance, but the other hand, it's very hard to move quickly and experiment when you've got to go through that kind of rigmarole. So, yeah, I hear you on that. 

David: Well, so how you're like, just get started. Right. Um, for the internal efficiency side of this stuff, the internal production and experimentation, how do you think teams should really think about integrating AI into their process? Because. The challenge is that you have existing workflows, like you still have a job to do.

And so I think this is probably something a lot of internal teams are facing where the company and management and external pressure, you read LinkedIn, it's all about A. I., you read all the headlines. We have to do something, but I still have to do my job. And [00:19:00] how do I kind of start to experiment and do some of these things without disrupting my current workflows? 

Dan: I have pretty strong opinions about this. Um, I think there's a path A and a path B and I'll talk through each one. So a lot of what I see what company is doing are path A, which is let's take our existing process, our existing workflows. We're going to apply A. I. tooling, and it could just be as simple as like Gemini built into Google docs or, you know, Office 365 copilot, or it could be something a little bit more, um, bespoke or, uh, novel.

They're basically trying to optimize, uh, the current workflow and it ends up optimizing around the edges. I have yet to see examples where that has a very significant impact taking that approach. I think, in part because. It's, it's almost like, you know, that, old chestnut, the, however much time you're allocated is how long the work will take, even if it does drive some [00:20:00] efficiency, other things just kind of suck up oxygen in the room, right?

Whereas I think path B, which I'm a much greater proponent of, is you take a separate team and you say, Hey, we're going to blow up the way that we work. The end outcome and the end output is the same, but we're going to assume, call it an A. I. native way of working. And we're going to build it from the ground up and it's going to have a different set of constraints and a different set of parameters.

you almost, this is not the ideal analogy, I haven't come up with a better analogy. It's almost like you are, are putting a different box around the working process that is not really even possible to do.

Whether that's a certain time to market, whether that's a certain velocity or output or a certain quality that you're able to achieve with like a smaller team, like whatever that is, you can kind of like create that and iterate on it in a small little vacuum. And then once [00:21:00] you fine tune that, and it's really humming scale it across the rest of the organization.

David: So sort of like a tiger team approach, like start a little startup within the company. 

Dan: I think so, because it just lets you kind of question all the assumptions you have, right? Um, there's, okay, you can talk about doing something faster or quicker with less, less people, but nothing's faster than not doing the thing in the first place. And so even though we're talking to A. I., I get enough into a lot of my clients where we're looking at process and There's a bunch of things that are happening today that I'm sure there were good reasons why they're happening, but in the cold light of day, you're like, yeah, we can just ditch that.

Yeah,

David: yeah, that makes sense. It's, you kind of, um, have to carve out people that are really thinking through the new problem. 

We've even seen that with, uh, like Flutter adoption, right? Over, uh, over the years in terms of like companies that try to do it a little bit on the side versus actually carving off a team that really focuses on learning from it and truly [00:22:00] evaluating it.

Um, cause it's too easy just to get sucked back into your day to day and competing interests of what you're going after. and the truth is that every company has a potential, um, Competitors starting like today, that is the future threat and the company starting today has the benefit of starting from a clean slate. 

Dan: I think you have to have that mentality. Um, in an ideal world, you're like, cool, I'm an existing business. I've already got, you know, a brand reputation. I've got existing clients, you know, a lot of the things around it. and if you can bring A. I. native workflows and capabilities to bear, then that's how you stay on top of all the new entrants.

David: So give me a quick, uh, um, overview. if somebody were, was interested in, they're like, okay, yeah, I hear you, Dan. Right. Uh, and I, everyone's talking about this and I got to do this. if you were to put someone on this path, right? What's like a three or six month plan look like for them?

Like, how do you kind of like march them through a process to kind of start to do some of this stuff? 

Dan: [00:23:00] Yeah, so I feel like all the main decision makers need to get a little bit educated. There's lots of great courses online. Andrew Ng, um, has a lot of free courses that I highly recommend specifically for business leaders, um, around A. I. or bring in somebody like an aspiration to do a workshop for your leadership team.

And then I think you have to focus on, on developing a bit of a strategy, which is as simple as to say, what are those one to hide? Uh. And then once you have clarity on that, assign a team and go off and stand something up and experiment. Whether that means getting an off the shelf tool, whether that means using A.

I. functionality that's built into your existing tooling. I do think that team needs to, like we discussed, Kind of start with a clean slate rather than, you know, iterate off of the current process. but yeah, that, that, that to me is like the clear path to starting to, to unlock value.

David: Awesome. That's very cool. Well, let's [00:24:00] pivot just from a time perspective. I want to get back to, um, A. I. in the user experience. You have, um, obviously a lot of experience in your career working for, for major brands and in their products, digital products, their, their user experiences and stuff. How do you think A.

I. is going to change the user experience? Um, how are you already seeing it reshape digital products and UX and where do you see it all going? 

Dan: Yeah. I think there is a, a lot of change that we can expect. I think it's still very nascent. Like I said, when I look at most businesses, they're less focused on their digital external facing interfaces, but it'll come. I think it has to come. I think first of all, from a pure UX perspective, I think a chat interface in most instances is a bad experience.

because, and I've seen this just even rolling tools out internally, you face a similar like paradox of, oh, what do I type in? Where do I even begin? You know, like nobody wants to learn prompt engineering. Like [00:25:00] that's, uh, I think it's, if you have to take a course. In order to learn how to use your product, you've done something horribly wrong, right?

From a, from a pure usability UX challenge. I think there's a lot of lessons learned from, um, Alexa, actually, you know, if, if you remember, um, cause, cause to me, it was a similar product experience of open ended conversation, With the device, right? whether it was through voice or through typing and the early versions of Alexa, all the Echo devices were just these little dots, right?

That all it had was a bunch of microphones and speakers and future permutations of it, they started adding screens to them because I think the team realized rightly so that there's a lot of information and interaction models, the old UI way that are just more efficient and better. For certain use cases, then open ended, uh, voice or speech or what have you.

So I think there's going to be some, evolution of more traditional, what we think of as UX and [00:26:00] UI and chat and speech on top of it. Uh, I, I really like, uh, the closest one that I've seen, I feel like, um, is actually cursors interface. Where you got that kind of co pilot thing going, but the transition between the two feels really smooth.

I could see some permutation of that working for like e commerce, for example, like complex, um, high consideration purchases. yeah, but, but I think it's still evolving.

David: The chat thing is interesting because what I find about the chat interfaces is first off, the types of customer service experiences you're used to having with like a chat thing. Everyone hates those. No one likes to like go to a little website and chat with customer service. So you already have like a bad mental model of that, even though the experience could be better because the information could be better and faster.

But then also the other thing is like, The way we chat, the way we use iMessage to chat and communicate with our friends or family, or the way we use Slack. Like if I look at my Slack channels, they don't resemble my chat GPT conversations at all in terms [00:27:00] of format, right? They're usually much shorter, quick little snippets of thoughts and ideas.

And so it's interesting that this main thing we do with chat, It, doesn't really resemble any other types of chats we're having when you think about it, So that's a little bit strange to me.

Dan: and this is where I struggle with it. It starts to look like coding. Like when you look at like prompt engineering, you're like, okay, I'm going to give you a step by step reasoning, you know, chain to follow. And it's like, well, that's just writing pseudocode at a certain point. And, and, you know, a lawyer or an accountant shouldn't have to learn coding, you know, and engineering, um, uh, mindsets in order to use, uh, use LLMs.

I think the other facet, you know, we talk about Flutter and mobile. I think I'm expecting multimodal capabilities to really unlock a lot in a mobile context in particular. I don't know how, I don't know exactly how that's going to shape brand experiences, but it just feels so powerful to, in real time, have your camera on pointing at [00:28:00] whatever you're looking at, you know, or be able to just converse, and have that translate into and influence some sort of brand experience feels, feels really important and really powerful.

David: Yeah,

I think, um, when you think about A. I., We spend a lot of time talking about like how to engage with the ai, like how do I talk to the LLM or how do I trigger an image to be generated? Like, that's a lot of the UX question. I kind of feel like where we haven't even really scratched the surface is like, how does AI actually dynamically influence UI so that the UI and user experience you're having is, is, is not even a, an AI experience, it's just that somehow AI is like influencing what you're doing.

Dan: I want, you know what I want? I want like the Matrix, that scene in the Matrix where he's like I need guns, and it's like, whoosh, like, the, the rack of all the guns, you know, come through, like, I'm imagining some version of, like, e commerce thing, where, like, show me tents, no, no, no, I want sleeping bags, 

just, just shows you, okay, now I just want the red sleeping bags, and, like, it feels, something like that [00:29:00] feels like it should happen at 

David: Yeah. I mean, some of the better experiences I've had with A. I. recently are around sort of this new version of search. Where, well, like for instance, we just added Slack A. I. to our org. I resisted it for a long time. I still have a lot of comments I'll reserve for, for another time. I don't want to, I will say that it's, um, not necessarily making me feel super confident about A.

I. investments, but, um, You know, the search is interesting, because a lot of search in the past was just like looking up matches, essentially, right? Like, I'm gonna go search my database for, uh, an instance of these characters in a row, or something close. Fuzzy search, things like that. The search stuff is cool, because it'll try you can like, ask it a more open ended question, and it'll try to go back go find stuff, and like, pull it together in an interesting way.

I think that's an interesting novel. thing we're seeing. And, and to your point.

of like searching for products on Amazon one day, and you're just like, okay, yeah, I need a, I need a sleeping bag. And it's like, no, no, no, I need a kid's size sleeping bag. And just a way to kind of like, uh, search through stuff.

Um, I think that seems pretty, pretty [00:30:00] compelling or like, you know, uh, even being able to parse it. Give me certain sleeping bags where, you know, people with really long legs don't complain about it in the reviews or something. Like I could see 

Dan: Something that is too nuanced to fit in like a traditional facet in nav, right? But you know, exists somewhere in the review data or something like, 

David: Yeah. 

Dan: um, it's definitely, yeah, that, that would be a dream for sure.

David: Are there things that you've observed or you're feeling in terms of like A. I. driven like interactions and UX like that? What are, are there other areas that we're just scratching the surface that are like big, unsolvable problems right now that we're trying to work through? 

Dan: Oh, man. Okay. So there is one that's been lurking in the back of my head that I keep encountering.

It's around this idea that we as users of technology have. Everybody is used to this deterministic software world. When I click the button, this is what happens. And if I don't click the button, that's what happens. And just, A. I. fundamentally doesn't work that way, right? It's not deterministic. and and [00:31:00] I think there is something around user expectations and managing those user expectations that we have not figured out yet.

and I don't know what the solve is, whether it's just like people get so accustomed to it, they're used to seeing either hallucinations or, Hey, I'm okay with this thing being correct 95 percent of the time. Um, but right now expectations are if, if it's on my laptop and it's giving me something, I expect that to be accurate a hundred percent of the time, um, or exactly what I want to see.

my, uh, oldest daughter is 11 and she's writing this book for school. And she's trying to generate, using, um, Dali, an image of the main character, right? And she's like, no, no, no, she has pigtails. And I could see her like yelling at Dali to be like, no, that's not what I meant. It should be like one big braid, not two small braids.

And there's this, you can just see the struggle with like, I have in my head what I want, and this isn't doing what I want. And again, with deterministic software, you would either say, yep, it does that or no, it doesn't. It's [00:32:00] very black and white. I know what to expect. Whereas I feel like with, with A. I., it's a bit, it's, it's more gray by definition.

David: Yeah.

you're right. We get like a toolkit now in our apps. Like if you think about using a Figma or something like that, you go and there's like a paintbrush tool and a color picker and stuff like that. But now it switches to like, it's not, you're not limited to the tools. Cause you can kind of just ask it to do whatever you want.

I do think the kids thing is interesting. You know, for me, uh, watching my kids use chat GPT. For them, it's so natural and it's almost like, uh, they're like, I'm like, isn't this amazing? They're like, no, I just expect that this would exist, you know, it's just seems very natural and obvious to them. Whereas we're all like, this is amazing.

Like what you're doing. and I also think the interesting fun thing about is like the, the shortening the loop of like an idea to something kind of real. Like my daughter followed one of the prompts that chat GPT gives you. And it was just like, I want to open like a sourdough bakery. And she, we found it like a couple of days later and I was like, Oh my God, what happened?

And she like made a bakery, like she made a menu. She like did the branding. She made a 3d [00:33:00] rendering of it like on the street, uh, this little town near our house. And it was just really amazing to see that creative energy. Like you could take what normally would have been just a very service level idea, probably never would have followed up on it or done anything.

And within five minutes, she had like kind of seen a vision come to life. It's just amazing to think that the acceleration of creativity and ideas that will come from that. 

Dan: Which is, which is really, really cool, right? I think that's, um, so cool to unfold. we're not opening up bakeries in this household quite yet, so have to chat with my daughter about it.

David: Well, in that case, uh, it.

was just ChatGPT's prompt, right? Like one of the, and that's interesting too, because then you can very quickly go down a rabbit hole of something totally unexpected. And you have this sort of, uh, this helper that can kind of push your thinking, which is really cool.

I think for the kids, it's going to be, I mean, for them, this is all just going to be very natural. So, uh, that's a good question though. Like, let's, let's think about that from a, uh, some of these kids who are going to be growing up. They're going to get the benefit where, you know, Five years from now, 10 years from now, [00:34:00] a lot of this stuff's going to get figured out, or the titans will have been sort of established, let's say.

I mean, you can maybe even say some of them already have, but what's your advice to companies around picking some of these technologies right now? Um, Like, how do you evaluate? Do you just do chat GPT? Do you do Gemini? Do you Anthropic? Like, do you do them all? Um, what's when it, when it comes to generative AI, I guess, specifically, what's your guidance for helping people pick? 

Dan: I think what you're going to start to see, I actually just published a piece on this. I probably should have elaborated this more, but like the actual foundation models are getting commoditized really fast, right? The prices are dropping, like in the last year, you know, down 30x, something outrageous. Um, and they're going to continue to fall and approach the price of electricity.

and so what I think you're going to start to see is the different providers start to differentiate themselves, just like any brand would. And, uh, you know, Around certain vectors that should make it a little more obvious. Which one is [00:35:00] right for your particular use case. So, for example, even today, there's only one model that will have a 2 million, you know, token context window.

Therefore, you can jam large videos into it, and that's Gemini, right? So, That mystery shopping use case I was talking about, like, uh, doing video annotation, it, it becomes Gemini is a clear choice, for coding applications. I feel like Claude has already kind of made a name for itself and in part due to its partnership with cursor.

Right? So, so I think you're going to see. A further pushing apart of these different models or for specific use cases, um, cloud isn't connected to the internet, um, by default, whereas chat GPT is right. So for applications where you do want Internet search baked into it, you want to deal with, like, calling a search A.

P. I. separately. You just want to baked in. Then yeah, chat GPT is great. So, I think over time, you're going to see, see more of that.

David: So some consolidation around just the standards and the models will ultimately become somewhat consistent and standardized. It's really the layers on top. [00:36:00] 

What about, um, uh, some of the non technical considerations people face with some of this stuff? Um, we get a lot of pushback, or there's a lot of fear, right, from A.

I., and privacy, you know, um, leakage of confidential information. Uh, we've had some clients that, you know, say, Hey, you're not allowed to use any A. I. tools because we're just not sure. There's the whole, like, lawyer side of it. The, the legal teams are not, not quite sure what's going on. how, how do companies navigate that? 

Dan: I don't think there is a silver bullet beyond what you just described. I think, uhm, You know, a lot of the different providers, they'll have, you know, pretty clear now terms of use, you know, associated with different tiers of payment. So, if you have an enterprise license, you get, uh, much, much better terms and have liability covered.

Um, I do think it just comes down to risk tolerance and liability, and I think it's going to be 1 of those things that evolves over time as everybody [00:37:00] collectively gets more comfortable with it. Uh, just. And like the broader public consciousness, um, that then company comfort level, we'll go up with it. I, there is however, like different, um, I spent a lot of time talking to lawyers about this, surprisingly, I, I got into this cause I thought the engineering was cool.

Then I ended up talking to a lot of lawyers. and it is also by application and vertical, right? So for example, in coding applications. You know, in part due to, I think, a culture of open source and like people putting their stuff up publicly on GitHub or Stack Overflow or what have you. It's not a very like litigious environment.

It's not a very litigious use case. So if you actually look at the risk of like using A. I. for coding applications, it's quite low. However, if you go over and you look at, um, um, art generation and image generation, uh, the visual art community historically has been much [00:38:00] more litigious than the engineering community, right?

So if you're going to start using A. I. generated images, a good lawyer would indicate that there's more risk associated with that. Doesn't mean you can't use it, but you should just be aware of the risk that you're taking on. So if you look at Firefly, I think half their value prop, Adobe's value prop for Firefly image generation is a pure liability play.

David: Hmm. Interesting. Just eliminating that risk for people. 

Dan: exactly. Yep.

David: Wow. Cool. Well, let's, let's wrap this up a little bit. And, uh, one of the things I want to do is connect the dots in your experience now that you're kind of focused on A. I. and where things are going and having, some amount of time behind You in terms of helping companies develop digital product experiences all the way from the Motorola, right, uh, through the, the, the mobile boom and everything else there.

When you look back on like what connect or what helped companies to be successful in the prior eras, now we're embarking on this new A. I. era. What are the things you think that are going to be consistent across teams and companies that are successful? [00:39:00] Moving forward, um, based on your experience in the past, like the, the sort of underlying attributes that help companies succeed.

Dan: You know, on the one hand it's really different, I think, but on the other hand, I think it's. And there's a lot of differences that we talked about just now, you know, expectations and the nuance of it. But in terms of leveraging it successfully, it feels like a lot, it feels very similar to the previous rounds of digital transformation, right?

Like you need teams that are inherently curious about the new thing. Um, you need teams that are, that thrive in ambiguity and are willing to embrace something new, um, and try new things. Which is easier said than done, um, but I think that's true here. And then the third piece of it is like having this clear focus on either the end user or the end outcome that you're going for and not getting caught up in kind of shiny object.

Um, I think those things still hold true. Uh, the, the nuances of how you [00:40:00] deploy A. I. and how the technology works, I know it's not deterministic. Like, yes, there's some slight differences there, but the philosophy and how a team should approach it. I don't think. Is any different than previous waves of technology.

David: There's always seems to be a little bit more fear and existential crisis to this one than past waves. I feel like, you know, if you think about it, what you're saying is totally true because we went through the first. com, right? You have all these, you know, brick and mortar stores and. You know, uh, uh, like figuring out how to get online, sell their products online, and then you have the mobile boom and everyone's got to figure that you got cloud stuff where you got to figure out how to move your data center to the cloud and all these different waves of transformation.

This one, the A. I. one seems to have a little bit more fear of like just jobs being eliminated. Companies just going out of business overnight because A. I. just makes them irrelevant. When you think about it, the New York Times is still around and people aren't really reading newspapers that much, you know what I mean? 

Dan: okay, but hold on. If you take a step back, I agree with you. And that's created a lot of hesitancy to embrace A. I. [00:41:00] But, the irony is, if you truly believe that, that A. I. is going to make a bunch of things obsolete. You would be in an even bigger rush to embrace it, if you think about it, not then, then, then the other way around, right?

Like, you should have even more urgency about it than somebody who doesn't have that concern. that, that, that's, there, there's a bit of a dichotomy there, I think.

David: Hmm. Yeah, that's a good point. Yeah. Well, so to that end, um, as just a sort of parting thought, like for an executive out there, you've already given us some advice on this front. Um, but maybe there's a company who's been dragging their feet and hasn't really quite figured out a way to, to get started.

What's your sort of like quickest advice to get them going? What would you say they should do first? 

Dan: I mean, I think everybody already has this, but if you haven't already, just use it for yourself. That's like the simplest starting point, is just to try it out for yourself and find one or two small use cases to get it going. And then if you're past that point, get your team to do it. And if you're past that point, then start thinking about, um, strategic [00:42:00] implications and, and high value use cases.

we're getting old enough where I think about like fitness, you know, and, I've done this to myself many times where I'm like, all right, I'm going to go out and train and I go from zero to 60 and then I injure myself. You know, I think a lot of the smart times just, just go out and do something.

Run half a mile, run one mile, do 10 pushups. I don't know, do something very simple. It'll build on itself over time. And this technology is not going away. You know, it's, it's, you're not going to nail it out of the park. It's going to continue to evolve, but I think it's more of the habit of starting to use it and, and getting comfortable with it, that matters more than anything else.

David: That's very good advice. Well, thanks, Dan. Um, this has been awesome. If, uh, anybody out there listens to this and wants to find out more about you and what you're doing, where can they find you? 

Dan: Eskridge. co, E S K Ridge. co online. Um, and find me on LinkedIn. 

David: Well, there will be some, uh, links in the comments. So, thanks a ton, Dan. It was nice having you on here. [00:43:00] Learned a lot today. And, uh, thanks for unpacking what is really complicated for a lot of people right now. and there's a lot more to learn. So, maybe we'll have you on, uh, a year from now when 

Dan: Oh yeah. And we can talk about what's changed. Yeah, sounds good. 

David: Alright