Podcast Transcript

[00:00:00] Sean Steele: G’day everyone, and welcome to the ScaleUps Podcast where we help first time Founders learn the secrets of scaling so they can fulfill a potential of their businesses, make bigger decisions with creative confidence and maximise the value and impact they can create in the world. I'm your host, Sean Steele, and today my guest is Mike Trigg. Welcome to the show, Mike. How are you today?

[00:00:16] Mike Trigg: Thanks, Sean. Great. Happy to be here.

[00:00:20] Sean Steele: Beautiful to see you. You are a 25-year veteran. That makes you sound older than you probably look, but you've been, not just a veteran of Silicon Valley is where I was going.

[00:00:32] Mike Trigg: so…

[00:00:34] Sean Steele: Yeah, me too. Me too. But you've been a Founder and executive and investor in lots of venture funded tech startups. You've been a contributor to TechCrunch and Entrepreneur and Fast Company and more recently, you are now an author of your debut book, Bit Flip, and you and I were introduced by Sean Flynn from the Silicon Valley podcast, and I was really interested by your book because whilst it's, and you know, hopefully I get the sort of do a reasonable job on the summary. And I fundamentally, it's kind of a corporate thriller, I love the way that you've written this, rather than just being like a kind of fact-based non-fiction. You know, it's a corporate thriller that sort exposes the underbelly of the tech industry in Silicon Valley and the moral dilemmas that the main character Sam Hughes has got to face, almost underpinned by this question, like, how far do you go to achieve your dreams of success and wealth. Which every Founder listening to this has been telling you, whether they're aware of it at the time, and they might just think of it as a difficult decision. We're facing moral dilemmas all the time when we're scaling our businesses, right? Like, you know, how far do I push that pricing? Or who do I hire in this circumstance? And is that actually discriminatory? But I really think they're the best person for all this sort of stuff. But you know, your book touches on important themes around artificial intelligence, as does your background and the ethical considerations that come with that. And there's no Seven-figure Founder, which is our typical audience who's not thinking about artificial intelligence today. You know, how they can leverage it, how it impacts their business or the businesses and lives of their customers, and as they grapple that, they're coming across all these ethical issues like, how do I think about privacy? What do I open up my, you know, is some AI tool going to integrate with my data set. And then what happens to that customer's data and do I still own it? And where is that now? And what about biases and discrimination? And who's writing the rules of the thing and if it's attaching itself to a data set that I own, do I have enough data actually for that to avoid biases or is it actually just going to be accelerating perpetuating biases that I've already got? And so given your, this is a big setup, isn't it?

[00:02:41] Mike Trigg: I don't know if I'm going to live up to the hype.

[00:02:44] Sean Steele: But given your last two kind of, you know “real jobs” before becoming an author, we're also focused around AI focused investment and incubation. I really today just wanted to kind of explore with you. Some potential AI use cases that people are thinking about, ethical decisions they might need to wrestle with. Your sort of thoughts and philosophies around, you know, kind of where this might be heading and how we think about it. If we're that kind of Founder, how does that sound like?

[00:03:07] Mike Trigg: Yeah, it sounds great. And it is rich soil to dive into. I mean, as you correctly state, AI has received a ton of attention not just by big enterprises and investors, but from down to local mom and pop shops are looking for ways that they can leverage AI to make their businesses more. And there's, because of that hype, there's also a lot of confusion around it. And I'm not surprised to hear that a lot of your listeners are looking to get up that learning curve as quickly as they can. So maybe just a little bit about my background, as you say, before becoming an author I've been, you know, as you stated in Silicon Valley a long time, doing various technology-based startups. Several of which leveraged AI technology. I was at a company that called Intello that does AI-based recruiting software. And then I did two rounds as kind of an entrepreneur in residence or advisor, maybe at AI focused investment in an Incubation funds. So one was called Symphony AI. it's about a billion-dollar fund. Focused on a automating various service services industries. So, a lot of what we did in researching potential ideas for new companies was to look at various service intensive, sectors from healthcare to shipping and logistics to agriculture and others. And figure out how different AI technologies could be applied to those to really automate a lot of the jobs. And that's in and itself, one of the ethical considerations in ai. Then I was at another one founded by a Stanford professor, Andrew Eng who's one of the more notable luminaries in the AI field. He created a fund called AI Fund, which is again incubating a number of companies and investing in other companies that are AI focused. So, I've definitely been in and around this space for a long time.

[00:05:20] Sean Steele: I think, jeez, you touched on a lot of things there and people will have, if anyone's been listening to the recent few episodes, we interviewed, Ryan Robertson from Bitten, who's actually one of my clients and he's scaling a national pest control business. And I was surprised to learn that actually all of their scheduling, because they've got technicians all over the place, you know, going to many appointments in a day. And so, they use a very powerful AI scheduling tool to make sure they take the right job with the shortest route, but that it's safe, that it's judging traffic problems. And so, that actually, that AI scheduling capability is a material advantage for them in their ability to help people be more productive, to have their technicians not spend inordinate amount of time in traffic, which is not good for their safety, not good for how they turn up to the next customer. Like it's a million different benefits to it

[00:06:13] Mike Trigg: I've been amazed at how many small businesses, services businesses do seem to be using some sort of technology. You know, whether it's “AI technically”, or just automation or chat technologies or other kinds of things, but I've seen it here in Silicon Valley, everything from a painter to a mobile dog grooming business that would have, you know, automated outreach to me via text message saying when their vehicle was going to be arriving to do the work. Automating their billing and payment process so that I get an invoice into my Email or text and can kind of pay that with one click. And so, a lot of those technologies you're seeing used by businesses that never would've been using the that kind of technology even just a few years ago.

[00:07:03] Sean Steele: And it's certainly, you know, you, you are comment upfront about how the fact it's kind of permeating, the business ward right now of course because of the advent of Chat GPT and, you know, there's been no faster user growth in any software to date, to my understanding, than Chat GPT and I was one of the first people to jump on plus since I've, you know, I'm excited, mucking around with GPT four on a constant basis. And of course, I have teenagers, and so my teenagers were straight onto it and we had to have lots of conversations and we're still having conversations about how this isn't a replacement for your thinking and how it's an enabler and your schools are all over this thing and they're looking for ways to catch you kind of plagiarising and all the rest. But so if we talk about maybe some of those different use cases. Actually, can we talk about where natural language processing is kind of heading? Like with things like Chat GPT, because it's on everybody's mind, right? So, if I think about, it feels like, correct me. I'm interested in your perspective. It feels like the next internet, right? Like it feels like such a big change, such a big acceptance of something that people probably found a bit scary before and all of a sudden have rapidly moved to because it's so easy and it feels comfortable for people, but I can only imagine that that means very rapidly we're going to have a sort of AI co-pilot, if you like, in all of the apps that we're using on a day-to-day basis. I would assume that we're going to be very quickly having, you know, a Chat GPT prompted response to the majority of our emails in Outlook that, you know, in Zoom we'll end up with kind of real time coaching about maybe, what to say next, what's just been said, what's been inferred, what does that mean for body language? Talk to me about where you kind of see this kind of heading and particularly in this sort of natural language processing.

[00:08:52] Mike Trigg: Well, yeah, I think Chat GPT by Open AI really captured everyone's imagination, I think to some degree, right? It was one of the first tools that you could, as a layperson go play with and start to enter information, whether it was your teenager. I have teenagers as well who've played with it quite extensively, maybe too much in the academic sense, but when you see it for the first time, it's pretty transformational, right? I think a lot of the natural language processing stuff that had up to that point been available and accessible was more sort of built for enterprises, right? It was really kind of behind closed doors, behind a big license. It was hard to train, hard to use, and it was really meant more for kind of understanding internal documents and natural language versus externalising and doing the creation of that content, the generative, and the generative ai. And so, I think it has really deservedly gotten a lot of attention and a lot of excitement. And it is a huge change I think for content creators. A lot of small businesses in particular struggle with the challenge of content marketing and social media marketing, that can be very effective, but it's incredibly labour intensive. I can attest to that firsthand. You know, I go out and promote my book and I've got to put blogs up in different places and I've got to do social media posts and sometimes your creative well just runs dry. And you know, those tools and there's a bunch of others out there, Jasper ai, Copy.ai, that you can use to create content. Now, you know, to my ear and eye as an author, I look at it and it looks, you know, and a lot of teachers will tell you this too, you know, we can tell when it's generative ai. But it's pretty good as a first draft. And, you know, that's how I've started using it in my own promotional efforts around as an author. I'll generate a first draft or I'll have it kind of go do a lot of legwork to pull in information that I might want to cite in my blog posts. I'm an active blogger as well. So those are pretty exciting

[00:11:24] Sean Steele: Just on that, so do you, I mean, you obviously have the copy for your book, so do you somehow feed the copy, like the full text? I don’t know how many thousand words that is, but I'm sure it's a lot of thousands. Do you feed that text into Chat GPT and go, okay, well based on this text, give me some ideas about what I…

[00:11:44] Mike Trigg: I haven't done that yet, although I'm intrigued by it. Right? As an author, you sometimes get the classic, you know, writer's block. And I think that's where a generative AI tool can be helpful. You can kind of give it a prompt and it can help you maybe break through and maybe it doesn't write the perfect thing for you, but it gets your creative juices flowing and you can kind of iterate or edit it from there. I did for the publishing of Bit Flip actually when it got around to producing the audiobook, I used a tool whose name is escaping me at the moment, but where I uploaded the manuscript of the book and I uploaded the audio recording of the book and it was able to identify for me. Discrepancies basically between the written page and the audio track and that was incredibly, incredibly helpful. I mean, you think, you know, the audiobook, I think clocked in at nine hours. That's a lot of time to go listen to it all. But with this tool I could accelerate, you know, I could just jump to the next place where there was a gap that it noticed between the audio recording and the written word, and then it gave me tools for reconciling that I could say, oh no, you misunderstood. That's correct. Or I could make the edit, I could mark it for my narrator who recorded the audio to go back and rerecord sections. If they were incorrect. So, you know, tools like that are being used in the editing, proofreading parts of publishing. And you know, there's a lot of excitement about it in the publishing world and there's a lot of angst about it in the publishing world as there is in academia and other sectors.

[00:13:31] Sean Steele: Yeah. So, if you think about from a, let's say we've got a bunch of business owners listening to this thinking, yeah. Maybe they're starting the process of starting to use it with their marketing. They're using it for sort of ideation, maybe rather than kind of completed writing. What do you think, what are the risks here and what are some of the ethical or moral considerations that we have to think about as business owners? If we were to get a bit sort of overzealous, we're really thinking, okay, we want to double down on this and really, you know, turn it into something that's actually quite in, you know, maybe rather than just me, every now and again, like it's actually built into our content sort of generation system or workflow.

[00:14:07] Mike Trigg: I think a big and maybe less intuitive risk is really to read it, right? I mean, to trust it, right? I think that there's, I did a prompt right before we got on this call to sort of see if I might be able to use it for an upcoming blog post. and, you know, asked it to do something where I was asking the algorithms to basically have some discretion, right? I said, “What are the five best thriller novels, you know of the last year or whatever.” There's no right answer to that, right? That's a lot of judgment and discretion and opinion that's baked into that. And so I think that's where you want to make sure that what you're creating and the content that you're putting your name on, ultimately represents something that you truly believe in and you're not just saying generate and you know, kind of off you go. The other area that there's a lot of ethical considerations and concern is obviously in copyright and IP – Intellectual Property. So, you know, there isn't great transparency in the tools as they stand today. I know this is an area that all the companies we've touched on are looking to improve. But where did that content come from, right? You know, is it truly just repurposed is how much is excerpted. And especially when it comes to those subtle things like, analysis and opinion where, whose opinion is it that's being echoed into, or summarised into this new content, because ultimately what it's doing is just generating content from the content it digests. So, like any computer program, it's only as good as the data that goes into it. And I think that'll be an area where you'll see some of the AI tools really improve, is to provide citations, provide better transparency into how that content is being.

[00:16:26] Sean Steele: And also, that's such an interesting point because if you think about that principle of saying, well, if I get Chat GPT or you know, some other kind of NLP tool to give me it's assessment, which is based on all its available data that's scoured from the internet, at a point in time we know that Chat GPT at the moment is sort of the data cut to my understanding is like September 21 or something like that. So, it's also still like two years, 18 months old. So be careful if you're relying upon it for anything that's up to date and accurate. Because I've run a few tests on that to you look for examples of different things where I've been writing a blog and go, gimme some examples of companies that kind of have these characteristics and 9 out of 10 have been completely incorrect or the company's changed direction, or the data's just totally wrong. But I guess more to the point, if you don't have a distinctive opinion, then essentially what you're doing is you're kind of almost by default becoming more generic because you're getting everybody else's ideas and then you're synthesising them down to something that's probably ends up being a generic version of everybody else's distinctive views, and you become just a generic producer of stuff. That's a bit of a brand risk. Like, you know, what do you want to stand for? Who do you want to be known as? And what does this all mean for trust? Because content is one thing. And I feel like we're at this pivot point, because we're so saturated with like there still seems to be, you know, every marketer that I talk to, I still get two different opinions. I get some who are still saying; the rhythm is the key. Like it's got to be weekly or if it's weekly, it's got to be every single week and you can't miss a beat. And it's, you know, it's got to be all predictable. And then other people are going, forget about all that. It's just about quality. Like, if you had to do it like once every se, as long as it’s good quality and it gets cut through and so on. That's far more important than volume and consistency, and I feel like we're at this sort of intersection where because we're so overwhelmed with content all the time, that quality lever really is going to have to go up. And so that's quite a risk if you start relying on Chat GPT.

[00:18:29] Mike Trigg: Oh, I think that's a better articulation of the first point I was trying to make is that, you're right, there's almost a dilutive effect to the process of generating content through ai. And the very real net effect of this, there're already. Is sort of more content created than we can consume on any topic. And you know, as these tools sort of facilitate more content creation, then the threshold to stand out from that sea of information is even harder, right? And if your process is to kind of take this distillation of an AI generated output and throw that out and expect it to get a lot of pickup and coverage, you know, that's already hard to do and it's probably going to just be harder and harder to do. You know, I think that's why I see these tools as sort of a starting off point that you can edit and iterate from. And, or inspiration, you know, giving you thoughts and ideas that you can maybe extrapolate from after that initial creative spark. But it's certainly not at the point and may get to the point faster than we think, but where you can just sort of rely on it fully. I actually think there's sort of an inverse thing that you're going to start to see AI tech technologies turn their attention to, which is consumption of all this content, right? I think I mentioned to you right before I moved recently. You know, I would need to buy a new microwave oven. What's the best microwave oven, right? Go, can I have a tool that's going to go out and not give me what advertisers want, but actually consume the content, read the ratings and the reviews and everything else, and make recommendations based on my personal needs and preferences and requirements. You know, that I think you're going to start to see companies address the other side of the equation too. It's not just content that's created and less and less of unless of it's being read, but tools for consumers and buyers at any level in, within companies and others to help them digest what's out there. And I actually think this is a really interesting, you know, a lot's been made out of Microsoft's investment in Open AI and how for fortunate that is and how scared Google is. I think for advertising-based businesses like Google and Facebook, this is potentially an existential threat. You know, it seems strange to say about some of the biggest companies in the world. But their entire model is to get our, get consumers in front of advertisers. Right? And I think AI technology opens the possibility of consumers actually funding what they want to find, rather than what a advertisers want them to find. And I think it's a threat to any ad-based business.

[00:21:22] Sean Steele: Yeah, and I think this is where this is where it's going to be a difficult place to navigate, right? Because what does Chat GPT, or you know, whatever the tool ends up being that you'll use to search for your microwave, or I need a project manager to run a construction project for me, or whatever it happens to be. What does the AI tool consider credible, you know, quality. How does it come to that determination? Is it because it sees a certain amount of volume? Is it because it's being like, you know, how capable is it in determining quality? So, we at the moment, we all sort of trust Google's ranking system to kind of bring us, you know, stuff off course that hits page one, we're all like, oh, that must have more credibility, blah, blah, blah. So, we end up sort of, you know, spending most of our time on page one. But what is Chat GPT, how does it think about, does it also go well, if it's on Google's, it's probably going to be uplifted in terms of its likelihood of being recommended by us, or is it, you know, completely even handed and, you know, to your point, I do a lot of video processing and audio and stuff on my computer, and I'd bought a laptop a couple of years ago and it was just starting to like grown under the weight of the amount of processing that I do. And I was really starting to get frustrated with just lag. And it was driving me crazy. So, I went to Chat GPT, and I was like, all right, I need a high, you know, I don't want a Mac. I'm familiar with Windows environments. I need a Windows base laptop that's got high processing speed for these kinds of use cases. And I used it, to your point, to get it to narrow down a significant number of options down to a few, give me tables, give me comparisons, and all the rest. I could then go out and check those out in the “real world”. but it's something that's really important I think for, as this shift starts to happen and as people start, like I am already training myself to, before I put something into Google, the moment I'm about to write it, I go straight to Chat GPT first, and I try to see if I can get a slightly more intelligent response than I was going to get in Google. And I think that's just going to continue to happen. And we're going to be forcing our team members to do the same. Like; Hey, before you search for any, before you actually start to write anything, start with something like Chat GPT. See if you can improve the quality of the question you were going to ask or the parameters that you give it. And so, you actually end up with less research time with a better quality response that you can then take and make some intelligent decisions around. But understanding how it's actually going to make those decisions will be really hard for us to know as business owners. Or what does that mean if we want to be one of the credible responses, how do we ensure that?

[00:23:55] Mike Trigg: Yeah. I mean, to that point, I've done something in the last month that I hadn't done in the last 10 years, which is used Bing to search for stuff, right? I mean, I like almost everyone, considered Bing to be a dead failed experiment and now all of a sudden, I will hit moments with Google where I just get frustrated. I find it not always useful for searches that required discretion, required judgment. And, you know, there have been so many, vendors that have optimise their content for SEO that you feel like you're just kind of getting regurgitated the same set of results. You know, that too many people have sort of exploited Google's tools to drive traffic their way. And I think, there's a lot of frustration from users about that sort of stuff and desire to find other.

[00:24:59] Sean Steele: If we put a heads back in the business owner space, talk to me about biases and how so, you know, I feel like one of the things that people will really be trying to, and I've spoken with lots of clients about this, you know, clients who have good size data sets, for example, you know, they've got 500 staff who are all doing a similar thing. Like how is the intelligence of the thing that they do being captured and is there a way for them to then having that decent size data set to do something with an AI tool that helps improve training, helps improve decision making, helps to kind of get the wisdom. If you think about most service-based businesses, it's usually based on selling people's kind of knowledge and time. And so how you capture that knowledge over time, how you capture the decision making processes, the way people are making intelligent judgment. How you might capture that on a dataset and then use some kind of an AI tool to be able to interrogate it in a more, a different way. But I'm really interested as to, let's assume that that's something that we can perhaps already do or is certainly going to continue to evolve. What does that mean from a biases perspective? Like talk to me a bit about how biases play out in that kind of scenario?

[00:26:08] Mike Trigg: Yeah. Well, you know, it's pretty simple at some level, right? The most ai systems learn by observing, right? They need the human input to say, this is the outcome we're looking for. This is good, this is bad. And that trains the model, how to replicate the human decision making that it's trying to automate. And so you're precisely right. Sometimes when data gets fed into a model, you can suddenly see that. You know, humans are, it's very difficult to make decisions without bias, right? Our amygdala, brains are wired to sort of make quick decisions sometimes for our survival instinct. If you think about it from a biological standpoint, you know, there's good reason over the millennia for why human brains and animal brains in general are wired a certain way. But you're absolutely right. You can certainly see bias and I for a while I mentioned at a company that used AI to automate the hiring process, and that is an area of fraught with potential bias, right? If you look at, okay.

[00:27:20] Sean Steele: Yeah. Can you give us some examples?

[00:27:21] Mike Trigg: Yeah, some of these were, were sort of famous, I think it was Amazon discovered they had an internal recruiting tool that they put together that was over-indexing to sort of white male applicants, essentially, right. And so, the tool that we developed in some ways helped eliminate that bias by obfuscating that information from decision makers, right? It would take things that could allow the human decision maker to introduce bias like names, ethnicity, religious background, whatever it might be and sort of anonymise that so that they'd really be evaluating a candidate based on their inherent traits and experience rather than any other factor. So, that's very interesting. You know, I think those kinds of decisions that are so, consequential for both the company and for the hire are a really interesting place to make sure that there isn't any bias being introduced. You know, I think there's also though, AI gets a little bit of a bad rap for the negative side of that, but it also can be a tool to ameliorate those, those issues, right? By feeding that data in, companies discover, oh gosh, we actually have some of these biases already. It isn't the AI fault. It's sort of baked into our hiring managers and other kinds of things, and so maybe we need to do a better job in our training and sensitivity training and diversity and inclusion training and other kinds of things to make sure that the human decision makers responsible for those business processes aren't, you know, inadvertently, perhaps even against their, you know, not consciously making bias decisions. So, AI can be a tool that can reinforce that bias, or it can be a tool that sort of shines light on that potential bias and is an instigator for change and improvement.

[00:29:33] Sean Steele: So, it sounds like, you know, the sort, the taking of the conscious, you know, like anything, right? It's the conscious application, as opposed to, oh, I've got some silver bullet that's going to fix everything for me and let's just let it, you know, kind of do its thing. How do I consciously apply this in a way that actually really gives thought to those things because they may be new. If you're a business where you've got, I don't know, 50 people and you've never really actually stopped to think about biases, full stop, you know, in your hiring process, but then you look around and you've got a whole bunch of homogenous, you know, like everybody's the same colour, they're all in the same age range. They all come from the same city and they, you know, and all of a sudden you go, wow, there’s not a lot of diversity going on here. How much of this is the available market versus the way that we hire, the way that we write the ads, to the way that we interview the clients, to, you know, and the bigger you get, the bigger that problem is, right? Because it's sort of, you can't see what's happening. So, I like that idea of being able to unpack actually what's happening before you then decide how you retrain or optimise for the…

[00:30:31] Mike Trigg: Yeah, and that's another interesting use case. I mean, another thing that this company did was had a tool for generating emails to applicants, to candidates. And it would flag language that was potentially, not discriminatory or overtly biased, but maybe subtly, off-putting to certain candidates. You know, one that's I was stuck in my mind was the use of the term rockstar. Like, we're looking for a rockstar. Well, that phrase didn't resonate as much with female candidates as it did with male candidates. And so little things like that can sneak into language of a job posting or an email and adversely impact the outcome.

[00:31:15] Sean Steele: Turn some candidates off.

[00:31:16] Mike Trigg: Yeah, exactly. And so I think that's another really interesting area for AI in more of the, as you get more into editing, is really to look at and flag for you potential, misuses and there are a lot of tools out there that do that sort of thing. And in my occupation as a writer, you'll see things that'll Grammarly and other kinds of tools that'll prompt you like; Hey, you're, you could, you know, word this sentence more efficiently, that aren't wrong per se, but there's maybe just a better way to phrase it.

[00:31:50] Sean Steele: What about privacy? You know, I imagine a circumstance where, let's assume we're using a tool. You know, we've, we've built a big data set, so we've got some strong kind of company IP that helps unpack, how we're behaving, how we're making decisions, and so on. We're then using some kind of an AI tool to make the most of that data set in helping optimise the efficiency of our business, whether it's in service delivery or operations, or how kind of how we do what we do. But then all of a sudden, we've got a whole bunch of information that might be sensitive and, you know, privacy of course is a big concern in every country. And the introduction of some pretty serious consequences for companies that don't take privacy seriously, I think has woken up certainly enterprise, I mean, enterprises really kind of all over it, but the smaller of town may be not, you know, not quite as conscious or not quite as literate for lots of reasons because doing a million things. But how do you see Privacy, sort of playing out. And you particularly in that kind of model where you may be leveraging a data set that's, that's yours inadvertently your customer's data, but all of a sudden, where does that data end up? And can you still…

[00:32:58] Mike Trigg: It's a very tangled subject for sure. This, we actually snagged on this in one of the pilot projects that we were doing for one of the AI firms I mentioned we were looking at, call data. We wanted to do an analysis of phone calls into customer service centres, and the client discovered they didn't really have the right to use those private conversations to train an AI model. And ultimately, you know, that project kind of got sideline because of that privacy consideration. You know, it's a real thing and I think that from a legal standpoint, companies need to start to maybe look a little bit into the future and figure out how they might need to use their own data and their customer's data for use within AI applications and write that into their terms of service, right? I mean, a small business needs to anticipate what they might need to do so that they aren't limited themselves. But of course, anytime you get into privacy, you also get into trust, right? You want to make sure that the people whose data you are capturing understand that you're capturing, understand how it's being used, have the ability to opt out. As you point out, there's privacy regulations all around the world, unique to geography, you know, unique in Europe, unique in California. So, it is a very, very tangled set of rules that you deal with, I think that's another reason, not to go back to the generative AI stuff, but I think that's another reason that those tools have been able to make as much progress as they have as quickly as they have, is that they're just sort of consuming the public internet, right. They don't face those sort of privacy concerns because the inputs are already in the public domain. And so I think that is a really important consideration for businesses. No customer or especially enterprise customer wants to feel like their data is being used for the benefit of other customers of that company. You think about a company like salesforce.com, right? I mean, they have arch rivals who are both customers of theirs, right? And so, they need absolute barriers between those different businesses as they develop AI. So, I think, you know, one of the things that you'll see more and more of. A lot of companies, especially consumer internet companies that were dealing with sensitive user data, had Chief Privacy Officers, and I think you'll see more and more kind of Chief Ethical Officers or Chief AI Officers who are there to, you know, make sure that the business is abiding by the law to start with, but also, I implementing best practices for how they manage that data and how they utilise it for training AI models.

[00:36:07] Sean Steele: It feels like there's going to be some emerging roles, right? Like, you know, people of, and whether they're outsourced roles and they're kind of outsourced to sort of consultants and so on. But, so one of course, you know, people are very concerned around cybersecurity as they should be. But this, you know, which feeds into this discussion because again, you know, you have a breach and all of a sudden your customer's data's all over the internet. To your point, that's a trust problem. So, cybersecurity might be the barrier between you and the trust problem. But this is also, how do you maintain trust? It's almost like the centre point is trust. You end up with this sort of, you know, trust team that includes cybersecurity knowledge. It includes data sovereignty and privacy knowledge. It includes, you know, maybe appropriate ethical uses of AI. Do you end up with some kind of a small subcommittee or advisors on your advisory board or something like that, that have got some of these specialisations? Maybe not all of them, but it feels like, I think, you know, one of the things that's really jumping out to me from this conversation is that the starting point is how do we ensure that we maintain trust? Whether customers, whether we're using it for marketing, whether it's about protecting the data, whether it's about leveraging the data. You know, do we have we have rights to use the data if we're going to start implementing tools that are going to change the way that we make decisions or improve, hopefully, the way that we make decisions. How do we do this in a way where we maintain trust at all times because it's broken trust that causes the biggest problems. And certainly, in any kind of services business where people are buying you and the trust in you that your people are going to be able to do what you said they're going to be able to do, or your technology enabled people, then that's a real problem.

[00:37:44] Mike Trigg: Yeah., I agree. I mean this is only going to become more and more acute, and I think that this is an interesting question for a lot of businesses, especially small businesses is, does the application of AI that they're considering enhance trust with their customers or potentially diminish it, right? And I think there are some examples where for a smaller business, that human touch may still be what very much differentiates that customer experience. And so, for that type of business, you know, the focus should be on how do you leverage AI to enable a more effective customer interaction rather than to, you know, replace that customer interaction.

[00:38:33] Sean Steele: Yeah, I am so with you on that. And I think, you know, one of the, you know, pre Chat GPT you know, the biggest fear I think of people when it came in the population, the consumer base was, you know, there's some AI tools going to replace my job, yada, yada. And I always thought of this as like the whole point of technologies, like artificial intelligence is to give us capacity to be more human because there are things that we can do that a tool can't. But if we have expensive resources, spending a lot of time doing things that actually computers will be far better at. That's an efficient unit of resource. And what you are robbing the person of doing is actually being, having the opportunity to be a hero with customers because they're spending all this time doing administration and stuff that can be done with workflow tools and you know, doing the research, trying to make better decisions like how do we help them have more positive, more powerful, more engaging customer interactions, which is actually, to your point, that's where the magic is. That's what the AI tool, hopefully won't be able to do in the future because it requires that sort of intersection of judgment and experience and understanding and empathy and emotional intelligence and all these other things that is certainly a lot harder.

[00:39:41] Mike Trigg: Yeah, I'll give an example. Being on the consumer side, recently, I was trying to get my cable TV service going, and I won't name the name of the vendor, but their approach seemed to be “We will 100% automate this customer interaction through a chat bot. Or on phone, the person will have to look it up, right? They have to ask me for my account number, my phone number, for my social, my last name, my address, et cetera. That should be information that's right at their fingertips, right? And so, neither experience was ideal, right? The fully automated chat bot couldn't figure out what I needed, and the human interaction required a bunch of questions that should have been at their fingertips. And so, I really think that hybrid experience is probably the right solution for a lot of, especially customer facing interactions where that agent is able to tell you what you need, answer your question, anticipate maybe what you're calling about, but still provide it through a human interaction versus an automated interaction.

[00:40:55] Sean Steele: And you know, I reckon one of the best places to figure out what these things are, because I think, one of the things we were chatting about offline is, well, yeah, should I use, yes, there's AI tools that can do lots of things, but actually, should I be just making a human decision here or should I actually be looking for a tool? Like, is it actually appropriate, or warranted or valuable to use that tool? But one of the things that I think is absolutely illuminating as you scale your business, you're just going to get further and further away from the front line. And so, what you lose sight of is actually all of the inefficiencies, the missed opportunities in a conversation, the magic in the conversation, how customers are feeling. Just sit and listen to the people who are on the phone. Or go out with them in the field and just watch everything they do for a day. And you will be amazed at all of the low-hanging fruit of opportunities of stuff that could be automated, improved. Like yeah, we did this, geez, I remember doing this exercise almost 10 years ago in, we had a 70-seat call centre that we'd set up for an education business. And we did exactly that exercise and we realised that actually our team was spend, it was like almost hours a day leaving voicemails. And the voicemails were almost identical, and I was; “Why are we spending two hours a day leaving voicemails?” Because you know, customers obviously young, don't always pick up their phone. I was like, can't we just make, still do it in their voice, record an mp3 And then when the computer hears the busy signal or it's about to get a voicemail signal, it just leaves the voicemail and they get fed the next call. And that in and of itself made such a big impact. I mean, they enjoyed it because they didn't have to leave all these voicemails, which were really in reality, other than saying, “Hi John”. Everything else was the same. “Hey, you made an inquiry. Just get in touch with you, ask you some questions about …”, so we recorded an MP3 for all of them, but it was still felt personal enough in the way that it was delivered.

[00:42:46] Mike Trigg: Vastly more efficient. No doubt. Yeah, no, I am a big believer too, in addition to the hybrid experience that, you know. If you don't have normal intelligence for a business process, you can't add artificial intelligence to it, right? Like I ran a coo of a company where, one of my groups was our customer service department. And we would do what we called fly-alongs where you'd sit in and you'd listen in on the phone conversations that our agents would have with customers. It was remarkable to witness it from their perspective. They had to log into three different systems oftentimes because billing was in one area, service provisioning was in another system, the customer CRM database was in another system, and that was why the calls took so long, they were so grossly in inefficient, you know, so fixing some of those things really made our process a lot more efficient. And so, you know, for businesses that sort of face that, AI might be sort of a bridge too far right now. The first step might be oftentimes is to get your internal house in order, really figure out how you can run that business process, whatever it is. Whether we've touched on a few scheduling or customer service or content marketing or other kinds of things, get it to be a well-oiled machine. And then AI can be an accelerant to that. Sometimes AI tools can be a convenient layer over those systems though too, and I think that's where, I use the example of like a cable TV provider. They probably do have a lot of arcane systems that are too expensive to upgrade or overhaul. And so having an AI system that's able to see, okay, here's the customers who's calling based on the caller ID. Here's the issue they might be calling about because we know in our database, there was an outage in their area or whatever it might be, you know, to combine those and roll those up for an agent so that they can, you know, guess, an educated guess what the customer's calling about.

[00:44:56] Sean Steele: Yeah, absolutely. So, what I'm hearing there, and I think that's a hundred percent the case, right? You know, workflow automation and getting your data sort of in order first, like try to make it efficient for them. And then if there's an opportunity to accelerate or improve or enhance, or advance the way things are done, then look for tools that can do that. I'm conscious of the amount of time we've got left Mike and I'm really enjoying this conversation. Is there anything else that you think that you'd really like to leave people with or a hard question or something that I haven't asked you that you thinks have been missed from this conversation that would be valuable for Founders trying to scale?

[00:45:32] Mike Trigg: I think that AI is, you likened it to the start of the internet. I mean, it does feel like we're at a very big inflection point in terms of this technology and its potential to change and improve lives. There are going to be a lot of potential negative consequences and ethical considerations and things that we've talked about, but I think on the whole, it's going to be very transformative. And you've seen a lot of technologies, you know, back to the printing press and the cotton gin and others where people have said, you know, this is going to cost jobs, this is going to drive people out of business. But in fact, what usually happens is, as productivity increases up, job training improves, everybody's sort of up levels and everybody's sort of quality of life is improved, the business efficiency is improved. And so, I think it's wise to go into this brave new world with eyes wide open, but I think there's going to be a lot of exciting opportunities for businesses of all sizes.

[00:46:36] Sean Steele: I love that. Mike, thank you so much for spending your time with us. How would people get in touch with you or follow along with what you're doing, obviously, where would you direct them to?

[00:46:44] Mike Trigg: Direct them to my website. Yeah, website, miketrigg.com with two G’s is a great place to see my writing on a bunch of topics, including AI, but also my, uh, writing work and there's linked to my books there and videos on YouTube and all that sort of stuff. So that's a one stop shop for all things, Mike Trigg.

[00:47:11] Sean Steele: Beautiful. Thanks so much, Mike. Folks, if you enjoyed the show today, number one, please join me in thanking Mike for his time. If you enjoyed today, you could just do one of three things that would be a great way to thank Mike for sharing his time, you can hit the subscribe button on your podcast app so actually other people hear his episode. You can leave us a review or share Mike's episode with someone who you know would love it and maybe is thinking about AI and the use in their business at the moment, that would be the world to all of us here. Please join me in thanking Mike Trigg. You've been on the ScaleUps podcast today. I'm Sean Steel. Look forward to speaking with you again next week. Thanks again, Mike.