By Humans, for Humans: The Right Way To Integrate AI

Transcription – By Humans, for Humans: The Right Way To Integrate AI

(0:02 – 0:14) 

Welcome to the Proof Point Podcast, where we decode digital success one click at a time. We share key takeaways fueled by data and insights that your team can implement today to drive growth. Now, let’s get started.

(0:21 – 0:38) 

This is Stacie Porter-Bilger, your host for the Proof Point Podcast, where I feature B2B and D2C businesses and thought leaders, sharing marketing, data tactics, sales strategies and leadership insights that will kickstart your growth for this ever-changing digital space. This is brought to you by Proof Digital. Proof Digital is a strategic and creative performance marketing agency.

(0:42 – 0:54) 

We partner with companies to create data-fueled marketing sales funnels and overall growth strategies. Visit proofdigital.com to learn more. Before I get started, I always want to give a shout out to a good friend of mine, Dr. Jeremy Wise, who introduced me to our guest today.

(0:58 – 1:18) 

Jeremy is a rock star and a speaker and a leader in the podcasting world. He really focuses on helping entrepreneurs, marketing and professionals to build relationships and help companies grow. So, a shout out to Jeremy. You can visit rise25.com to learn more. Second, Jeremy is awesome. That’s awesome. He is. He is. He is a rock star.

(1:19 – 1:41) 

Our guest today is Ken McLoud. Ken has spent over 15 years as an engineer working with brands like Toyota, Lockheed, Martin, Bose before striking out on his own to build custom AI software for growing entrepreneurs. He believes that implementing new technology is a three-legged stool of strategy, implementation and follow-through.

(1:42 – 2:06) 

He brings all three onto every engagement. One of the pieces of feedback he’s most proud of, of his clients, is telling him, you didn’t build what you asked for, solving the actual problem rather than spec, and that’s why he continues to grow and help companies grow. That approach has produced real results, including one AI system that increased a client’s profit over by $20,000 a month.

(2:10 – 2:28) 

Ken runs Laconic Tech from the mountains of New Hampshire, where he lives with his wife, three kids, and a chocolate lab. Awesome. Who’s snoring away on her bed right over there. I know. Thanks for joining me. Gosh, I love that. Thanks for having me. Yeah, and you’ve got a lot of land for the lab to run around on. Absolutely.

(2:29 – 2:55) 

Because you need to do that with labs. Anyway. Well, thanks for joining me, and we’re both jumping in and two feet into AI and the magic of AI. I think the big thing here, what I really want to get into is how you can customize it and the beautiful thing, especially for entrepreneurs and small companies. But before we get into all the fun stuff, tell me a little bit about your journey. I did the bio, but I would like to know a little bit about your story.

(3:00 – 3:12) 

Yeah. So I’m an engineer by training, and I spent most of my career working regular nine to five corporate jobs, though was always sort of a entrepreneur in that phase. Always had one side hustle or another burden.

(3:17 – 3:44) 

The first real success I saw in that started in 2019, just before the pandemic, and it was doing custom software development for other small, often manufacturing companies there, where it would be like they couldn’t find a solution for what they wanted to do with off the shelf software. So I would work with them to build out custom, usually back office software to get it done. That was doing well on its own.

(3:46 – 4:12) 

And then when AI kind of exploded onto the scene around 2023, I just provided a whole new set of tools to supercharge those workflows. So all of the problems that we couldn’t nicely fit into if then statements for all the variable documents that you couldn’t come up with one parser to parse all the different kinds of documents, you now have this skeleton key to attack those problems with. So that started going really well and helping the business even more.

(4:17 – 4:36) 

And then shortly thereafter, I had survived three rounds of layoffs at my day job, and the fourth round got me. So I said, like, there’s there’s my signal to jump into this full time. And it’s been fantastic ever since. The layoff was the best thing that happened in my career. No, I tell you. And I mean, now you’re ahead of the game.

(4:39 – 5:23) 

I mean, I mean, you might I mean, it’s it’s gone. The train is off the station. But those who jump in and are going to win in the long run. So what are you what are you seeing? I mean, what you’ve seen from an engineer standpoint, what’s the practical impact? I mean, what is you know, what what is the impact of those who are struggling to apply it and those who are not? Yeah, well, the big thing here is thinking about these systems as giving our existing team superpowers. Yes. So like the people who are trying to think of it as wholesale replacements for humans and in nearly all cases are not having success, but these systems are just not there yet.

(5:30 – 5:56) 

Plus, boy, if you want a surefire ticket to have your team rebel against your AI efforts, like make it smell like you’re trying to replace the whole team with AI. Yeah, well, that’s not happening. The the real way to run these is to take, you know, let’s say we could handle three clients per employee before we look at what’s consuming their time, what the kind of low value tasks that are chewing up a bunch of their time are.

(5:59 – 6:17) 

And then we automate the snot out of those with AI so that maybe now you can do six or nine or 10 clients per employee, handle way more volume with the same team size. Right. You used the word earlier workflows, right? What are your workflows? What are you doing and how can you optimize that? Not necessarily, you know, a new shiny object here.

(6:20 – 6:38) 

You’re just talking about scale. Yep. No, that’s a great example. So if you’re thinking in terms of roles, you know, I have these five roles in my business. Which one can I replace with AI? It’s almost certainly none of them. So like a lot of times the bad taste some people have in their mouth will be that they tried to wholesale, replace a single role, and it’s the tech’s not there.

(6:43 – 7:06) 

However, each one of those roles can be broken up into various workflows that those people do. And sometimes the workflow will cross across multiple roles. And there are a lot of examples of workflows where we can automate, you know, either the entire workflow, or automate a lot of it with a human review step at a couple points along the way. Yes. And then be able to do way more volume with the existing team. Yeah.

(7:06 – 7:17) 

And that’s in the tech world called the human in the middle kind of thing. That’s the kind of the funny thing that we’re using. But it’s, it’s just really putting those systems in and having QA quality controls within it, just like you would in an engineering process.

(7:19 – 7:32) 

And then give it some information that it needs to make sure that it’s the output is, you know, what you want it to be. Any, any, before I get into, I want to get some case studies and some stories. Yep.

(7:33 – 8:04) 

But, you know, the piece that, that is an opportunity going forward, we talked a little bit, is, you know, we used to kind of look for off the shelf opportunities, right? We always kind of, from a solution standpoint, we did it. But now, you can do custom AI solutions at, I shouldn’t say cheap, but I mean, at where you had to do huge expenses as a company to get custom work. With AI as a solution and companies like yourself, I mean, I didn’t, I don’t want to diminish the value of it, but you can speed up the process of development with AI.

(8:08 – 8:30) 

Certainly. Yeah. So it’s the kind of stuff that a handful of years ago, you know, one of the tech giants would come in. Right. These big multi six figure proposals to enterprises for that were just like inaccessible to mid market or smaller companies that we can now approach for way more acceptable price tags. Right.

(8:31 – 8:41)

Yeah. I mean, small companies, medium sized companies didn’t have that option to use custom solutions. They had to make the off the shelf work within their systems and it was clunky and it didn’t work.

(8:41 – 8:52) 

It didn’t really solve their problem. Now, I mean, what you’re doing, what your company is doing is bringing those custom AI solutions to entrepreneurs and help them scale. Yeah, I know.

(8:53 – 9:10) 

And a lot of the times, one of the biggest costs of those off the shelf solutions is that they make you change your business process in order to fit the mold of that off the shelf solution. Yeah. A lot of times that can incur a lot more cost than the actual pay and the subscription fee of the software.

(9:10 – 9:20) 

Yeah, it is. And, and, and although, I mean, it was good at the time, I mean, it gave a little bit of a bridge so that you could do some things that maybe the big guys are doing now. I mean, we’re in a window right now.

(9:22 – 9:34) 

I mean, there is an advantage to me from a small to medium sized company of adoption if they do adopt. That’s the thing. I mean, they have to, they have to be, what are some things that you see of those who are saying, okay, AI, we need to look at it versus those who don’t.

(9:39 – 10:11) 

Anything that you see out there? Yeah, so a lot of it comes from the founder or the CEO, the current person who’s running that business. And like the, the most common pattern I see where I wind up getting involved is the situation where you have a techie founder, you know, the kind of person who’s buying the latest gadgets and playing with all the latest AI tools and then seeing them online, you know, they’re listening to tech podcasts, but they’re not, the business isn’t a software business, right? Cause then they wouldn’t need me. They’d be doing this stuff on their own.

(10:13 – 10:54) 

It’s a less technical business, like an agency or a manufacturing operation, a logistics company where they don’t have an internal software team, but the founder is using this stuff personally and sees the opportunity and wants to take advantage of that. The magic that can happen with it. We built some, just some competitive analysis tools and we kind of, human in the middle, but we, you know, did some simple, I mean, it’s not totally simple, but we’re scraping data and doing some things on and what it’s, I shouldn’t say spitting out, but what’s coming up with is crazy, good, crazy, good.

(10:56 – 11:11) 

And you know, we demonstrate that some clients in there, like who would spend thousands of dollars on a report like this. And so, we’ve got, we’ve got a whole kind of line of successful projects that I call AI powered lead magnets. Can you go into those? Yeah.

(11:11 – 11:29) 

Yep. So the idea is, and this can apply across a bunch of industries, but you probably have some expertise that your clients find valuable, right? Whether it’s about SEO or now GEO or about site design, conversion rate sort of stuff on a site. Maybe it’s more technical and is about ad performance or creative ads.

(11:35 – 12:08) 

Maybe it doesn’t have anything to do with that kind of tech marketing stuff. And it’s the sort of thing you can ask a questionnaire about. Yeah. If it’s about putting a new driveway in or a new roof on your house, the sort of thing you could ask a multiple choice series of questions about. So we, we gather this information from the prospect. We then send it off to an AI along with what do you can think of like an SOP document that you’d write in order to teach an intern how to evaluate this website or how to evaluate this new roofing job, et cetera.

(12:11 – 12:39) 

And then we can produce rather than a generic top 10 tips for re-roofing your house PDF, you can produce one customized specifically to their situation and deliver it to their inbox. So now you’ve led off with value, you know, here’s some tips that apply specifically to your situation. You’ve established yourself as an authority in that space, and you’ve come up with a convenient excuse for grabbing their email address so that you can reach back out to them later.

(12:43 – 13:08) 

Yeah. No. And that’s those simple inputs. And then you can spit out, again, I have to come up with a better word of an end product  that creates value and actually helpful. And it’s a, you can, from certain agencies, it could be a lead source and also could, you know, it could be a way for you to develop proposals or, you know, automate those types of things. And yet some other examples.

(13:10 – 13:21) 

I mean, I think a lot of people just want to know how are people building these custom solutions and what are they doing? Nope. So I’d say this, the second big category falls into that workflow process. So these are usually more agentic tools.

(13:24 – 13:41) 

And when I use that term, I mean, the AI kind of gets to decide how long it’s going to work for before it finishes the job and can decide to call different tools and take different steps. So rather than just an example like the grader, okay, AI, we’re sending you this batch of information. We want you to give us back, you know, the code that’ll generate this PDF, let’s say.

(13:46 – 13:55) 

Yeah. Content that we’re going to populate into a PDF template. With an agentic system, it’s, here’s a problem, here’s a list of tools that you can use to solve the problem.

(13:57 – 14:30) 

Now get to work and let me know when you’ve achieved the success outcome. So this is stuff like a great example from the intro there. You talked about the $28,000 a month result. So that’s a logistics company out of Memphis that has, so basically they are a trucking brokerage. So they get loads tendered to them from their customers, you know, this truckload of two by fours is going to go from Memphis to Tallahassee for a particular price. And then they can choose to take the load.

(14:33 – 14:55) 

And then obviously what they need to do is actually get it moved for less than that amount of money so that they can turn a profit. So they had way more opportunities to bid on these loads than they had time for people to actually do the due diligence on how much is it going to cost us in order to move this load and decide whether or not. Right.

(14:55 – 15:09) 

Do we want this? Do we want this deal? Do we not want this deal? We’re going to lose our shirt on this deal. We’re going to take this deal. And if it’s, you built that, I mean, this is a custom solution. It’s a, I mean, you have your agents. This is a little higher advanced. This is your automation.

(15:10 – 15:37) 

This is a little more advanced. This is actually a custom solution, leveraging AI to help you make decisions. Yep. So it’s automatically pulling these opportunities from the customers. It then both looks through a proprietary internal database, right? How much it’s take this, it’s cost this company to move freight along that lane, right? Also calling some external APIs to get things like what trucks are currently available, what kind of money they’re looking to get. Current diesel prices is a big deal.

(15:39 – 15:55) 

Right. I’m sure. Absolutely. I mean, yeah. Going through the roof. Yeah. And then the AI follows what is essentially an SOP document on how to determine a bid price for that job. Or perhaps there’s some, we just don’t want to bid at all because they’re losers. Yes.

(15:55 – 16:15) 

And then the AI goes out and places the bid on those jobs. So whereas before kind of these load boards would get looked at sort of intermittently as people had time and the vast majority of opportunities were just going on bid. Yeah. And now nearly every opportunity gets bid on. You win some significant percentage of them and. Right.

(16:15 – 16:32) 

And you’re making better decisions because you, I mean, the AI, you know, custom tool here is actually giving you real time data. You’re doing all the due diligence on 100% of the opportunities, which is not something that humans are generally doing. No, no, no.

(16:33 – 17:01) 

It’s not possible. I mean, I wish we were that smart or that fast. Yeah. But I mean, the opportunity and logistics, I mean, I see a huge opportunity in logistics as far as money made, money saved across the board. So I don’t know anybody working on a logistic company, I know who you need to talk to. Okay. That was a great example. A great example. Other, other, others you want to share? Yeah.

(17:02 – 17:36) 

So sort of the, the third example after these AI powered lead magnets and the workflows are AI data analysts. So there’s a lot of companies out there who are data rich, right? You’ve got a huge library of marketing data, of operations data, all sorts of stuff that it would be really nice if you had somebody on staff who is a real data geek and could jump into those databases and get you the answers that you want to run your business. But a lot of companies can’t justify having a full time, you know, this is definitely going to be a six figure hire of a data analyst person to answer those sorts of questions.

(17:42 – 18:10) 

So what we can do now is take one of these agentic systems like we were just talking about and wire them up to the company’s proprietary data sources. And now you have what looks like a chat interface, like you’re used to with chat GPT or Gemini or any of the others. But now it’s got access to your internal data sources and instructions on sort of how to interpret them, what they all mean, how they relate to each other, how we calculate our gross margin this way, calculate this metric that way, these kinds of instructions.

(18:15 – 18:36) 

So now the business leaders can go in and just like they would go to a human analyst and ask, you know, hey, I’ve got this big job coming up and I need a bid. We’ve done three similar jobs in the last year. Can you pull them and give me average cost on those jobs so I can use it to bid this new job? The agent just goes ahead and does that.

(18:40 – 18:57) 

I mean, when some small companies, means that companies are listening to this, the one thing that they’ll think about, okay, how do I keep my data secure a little bit on this? So once you get out and get it, it’s not the same thing. You’re not putting this in chat GPT. This is not the same thing.

(18:58 – 19:13) 

And actually it’s not a good idea sometimes to put that kind of data in there. Most of the time it’s not. So anything, anything that’s going into the public, you go to chat GPT, especially on a free account, they’re like very open about the fact that they’re taking that data and training on it.

(19:14 – 19:28) 

So it’s going to wind up in the public domain and it’s okay, sometimes it doesn’t matter. A lot of times it doesn’t matter. When you’re using one of the paid APIs, all of the providers are promising that they’re not training on any of the data that goes through the APIs.

(19:33 – 20:05) 

And then if you have super sensitive stuff, there are like, I know Anthropic does this, Gemini does this, you can pay extra and this is certainly a higher tier sort of thing to be running on a private instance of the AI within the cloud. This is how like when you hear like with the war going on right now and you hear about them using AI and people are getting in over it and you might think like, wait a minute, how the heck are they using it for sensitive military stuff? Like that’s how they’re doing it. It’s flowing through a dedicated compliance controlled instance.

(20:11 – 20:30) 

Yeah. Yeah. And I mean, like you said, if I mean, if you are using paid versions, and if you actually have, I mean, cloud, you can download into a, you know, desktop. Well, you’re not this is an important thing to understand here. When you download cloud desktop, the thinking is not happening on your desktop. That’s true.

(20:31 – 20:45) 

That’s a good point. Sorry. That’s a good point. That’s giving you an interface to it’s still sending the data up to the Anthropic. Gotcha. That’s a good point. Good point. There are AIs which can run locally on your computer. The downside is they are a lot dumber than the ones that we’re all used to.

(20:51 – 21:06) 

You know, talk about that. I mean, everybody always asks me this question, which which ones are, you know, whatever. I mean, to be truthful. I mean, it’s changing all the time. It’s not a I mean, right now I’m kind of in the I’ve really been impressed with Anthropic and Claude. But, you know, Gemini is doing better every week.

(21:12 – 21:27) 

You know, ChatGPT was the Kleenex of AI and I call them that because they were kind of the first out of the gate. But, you know, they got a lot of investments going to figure that out. We’ll see how that goes this year. But you know, they’re all they’re all under race. Yeah, no, 100%. So a big philosophy of how we build these systems is we try to avoid vendor lock in.

(21:33 – 21:45) 

Yeah. So like we don’t want to pick we’re using Grok for this and then just always run there. We want to architect the systems so that the LLM, the brain is an interchangeable Lego brick that we can unplug that one and plug the next one in really easily.

(21:50 – 22:18)

And we do that because all of these major labs are spending billions of dollars a year. And the there’s two things going on here. One, the one that’s best at any given time will change every few months. Like, I agree, anthropics on a tear right now. But like, would I bet much money that they’re going to be the leader in six months? Like, probably not, because it’s changed three times in the last six months. That’s right.

(22:19 – 22:26) 

That’s really important. You know, it’s you got to change out the blocks, you’re building the workflow, you’re building the systems. And again, this is a system you should work on fixing no matter what.

(22:29 – 22:45) 

I mean, you got to work at that. What is the workflows? What are you spending most time on? What do you need to save time on? How can you increase your bids? You know, what’s the problem you’re trying to solve? And then, you know, you know, pulling that breaker and not brick depending on what’s working today. Yep, absolutely.

(22:45 – 23:08) 

And then the other thing that happens with with treating them isn’t in interchangeable Lego bricks. That’s the one. Yeah. You got it. You got it. Is that certain LLMs are better at certain jobs. They are. Like we’ve mentioned Gemini a couple of times recently, and Gemini is just like hands down the best at interpreting anything visual. Yes.

(23:09 – 23:21)

Like I have some manufacturing clients where we want the AI to be able to understand an engineering drawing in order to make some decisions. And we’ve all of the major models can take images as inputs. Yeah.

(23:21 – 23:31) 

When we give it something like a drawing and then ask it questions about it, Gemini is just head and shoulders, like not even last month, the last month. Crazy good. Yeah.

(23:33 – 24:07) 

Then we also run into things like we’ll have we’ll often have a case where you have this enormous pile of data and you’ve got to look through all of it because you’re looking for some needle in that haystack. But the problem is the frontier models, you know, like Opus from right from Anthropic or Gemini or the three one pro now from Google are pretty expensive to send huge piles of data through. Now, like anything you’re dealing with that you’re typing out on a prompt is negligible and don’t worry about it.

(24:08 – 24:27) 

But these are like, you know, we’re sending books worth of data through because we’re looking for some particular little thing. So if you’re doing that through the big brain model, it’s going to get expensive fast, which if it’s a one time thing, probably not a big deal. But if like if this is a workflow that you want to run every day, yeah, it’s going to matter.

(24:27 – 24:52) 

So we’re often looking for small, smaller, cheaper models that can act as a filter for the big brain. Right. So we don’t want to do our our heavy lifting decision making with one of these little models. But a lot of times we just need, you know, there’s a thousand documents here. We got to read through all of them and find the four that are actually important for the big brain document to go chew on. Right.

(24:52 – 25:05) 

And a lot of times those won’t necessarily come from the same provider. Right. Like right now, one of our favorite ones for doing that kind of work with is GPT mini. Oh, is that right? Open eyes. Yeah. It’s tends to be really good at following instructions.

(25:08 – 25:28) 

You know, if we want things that are that meet these criteria and not these criteria and the other smaller models like like Haiku, which the other problem with Haiku is it runs kind of expensive for being a small model. Yeah. And flashlight from Gemini, a lot of times we’ll run head to head comparisons and GPT mini will beat them.

(25:29 – 25:47) 

And that’s what you got to keep an eye on. Right. And to your point, I mean, you know, the number of calls, the number of integrations that you have, the amount of data, all that, you know, could could kill the cost of actually making it worth. Yep. Absolutely.

(25:47 – 26:08) 

So the it’s important to understand how the pricing on this stuff works, because we often won’t be able to tell you upfront it’s going to cost X dollars per run. Yeah. Because essentially you can think of it as words of input and words of output. Yeah. You pay a certain price per million words of input and a certain price per million words of output. That’s an oversimplification, but it more or less works.

(26:12 – 26:29) 

So for a particular workflow, it might be that on on Tuesday you go through one million words in order to get it done, and on Thursday you go through four million words to get it done. Right. Like if it’s a thing where we’re cost sensitive, we can build controls in right and make sure that it doesn’t ever exceed X and it’s not going to run away.

(26:32 – 26:47) 

Right. But it’s it’s usually not something we can predict well up front. And if you’re frequently going through lots of data, it becomes like the number one thing that you’ve got to engineer around to make sure that you’re not spending any tokens you don’t have to.

(26:48 – 27:08) 

Right. And as we get into this new word of, you know, language of tokens and calls and tokens, I mean, that’s what you basically buy. Oftentimes you get in there, you go in there and we like I said, we were doing this competitive intelligence tool and build it across multiple systems and, you know, OK, is it worth that? OK, we need to get the price down on that amount of tokens we’re using.

(27:11 – 27:35) 

How can we do that? What do we have? We look at it from how it’s engineered, talking to my engineer here. So, you know, it’s a process. But I think understanding the workflows, understanding a bunch of data and testing that and then coming up with, OK, is it worth it here? Do you need this? And is it OK without this piece? And does it still solve the problems? The first part of almost all of these projects looks way more like business consulting.

(27:40 – 27:53) 

Yes. It’s like tech. Right. It’s about defining, like, what is the process for achieving this goal? A lot of times it’s not well defined and we’re going to try to boil it down to a list of steps. Right. Right.

(27:54 – 28:21) 

So it looks a lot more like management consulting than it does like tech a lot of times at the beginning. It doesn’t. But it should. We’re talking about business systems here. We’re talking about how every company should be looking at how can they be more efficient and effective in their business? What’s overloading this part of your business? Where’s some efficiency to be had? But make your team, back to the piece, part of the solution. I think that’s what you meant that early on.

(28:24 – 28:46) 

It’s not a without a human in the middle here. We’re just trying to scale and actually create a better, stronger organization for everybody involved in our company. And to pretend that, to close our eyes and say, it’s not going to go away, it’s not. It’s going to go through some hiccups. I mean, there’s no question. I mean, I think there could be, I mean, we all, everybody is going to be a bubble or not a bubble.

(28:47 – 29:03) 

Yeah, maybe. But that happened before Google came to be and look where Google is now. So I mean, you know, search was the same thing. It’s been dead for two decades and it’s still going strong. Yeah, absolutely. So like, might some of these stocks collapse? Sure.

(29:04 – 29:24) 

But do I think there’s any significant chance that five years from now we’re going like, oh yeah, LLMs were a fad. Nobody uses LLMs to do business stuff anymore, like, no. It is, I mean, I don’t want to, it is the most significant shift in how we are going to work and live.

(29:25 – 29:36) 

I mean, I don’t mean to make this like a, you know, but it’s truth than the industrial time period. I mean, I think it’s that big and it’s going to change a lot of things. It’s going to be a wild ride for sure.

(29:38 – 29:58) 

It’s going to be a wild, wild ride. And you know, I’m choosing, there’s concerns and there’s no questions, but I’m choosing to know and understand and be an advocate for how can we help companies be more secure with their data? How can they think through this strategically so that they can grow? But yeah, it’s going to be a wild ride. Yeah, absolutely.

(29:59 – 30:35) 

No, that’s a good segue too into, a lot of times the concern in picking a project is that we want to really avoid being a solution in search of a problem, right? So like the, you’ll get this thing where a business leader walks into the room hyped off of a conversation like the one you just had, and they’ll go like, we’ve got to do AI. We’ve got to do more AI. And like this way almost never leads to positive ROIs and good outcomes because there’s the reason why like solution in search of a problem is a cliche, right? That’s a crap in human thinking.

(30:37 – 31:08) 

And instead what you want to do is, is put that business consultant hat on. Think about what are the constraints in the business that are keeping us from achieving our goals? Right? Maybe it’s revenue growth or profitability or some other internal metric that we’re tracking. What’s keeping us from getting there? And then once we identify that, look at it through the lens of these new tools, because there’s probably relevant tools to that problem that we didn’t have six months or a year ago when you considered that problem.

(31:10 – 31:31) 

Right. And if you use these new tools to attack that problem, you’re going to get a hundred times better ROI out of it than if you just go, we’ve got to do AI. Somebody find something to do with AI here.

(31:23 – 31:40) 

Yeah, no, that’s, yeah, that’s what, yeah, it goes back to, it’s still business practices. I mean, continue, I mean, especially you come from a manufacturing base. I mean, you know, there are, I’m trying to think of the best book. The goal is a long time ago, it was a book. The goal. I mean, you remember that book? I mean, it was, we mentioned earlier in the podcast has me reading it.

(31:45 – 32:01) 

Yeah. I’m like halfway through right now. Yeah. I mean, it, the, the, it’s an old book. I love the book, but I mean, it’s like, it’s written as a novel too. It’s like, I guess you’d call it like a parable or it is a parable. Yeah. Yeah. Yeah.

(32:02 – 32:11) 

But it’s about understanding. I mean, what, what, what are the blockers? What are the roadblocks within your businesses? We know where the, where is the problem? You think the problem is over here, but it’s not over there. It’s, it’s over here.

(32:13 – 32:32) 

And so it’s the, it’s the same thing from a business consulting standpoint. I think it’s important to, to reinforce that. I mean, I’m, I’m all about, you know, understand AI, go ahead and, you know, put a, you know, see what you can do as far as creating a letter or an email or those kinds of things we all kind of play with or, you know, pick a trip somewhere and what should I do? Those are all fun things.

(32:33 – 32:49) 

But here, here, here is an opportunity first to really look at your business and look for opportunities of where you can find improved workflows, improve efficiencies. So it’s a Jeremy’s got you working, reading that book again. I need to reread it. It’s a long time ago. It’s an old book. It’s not an old, it’s not a young book, but it’s called the 80s.

(32:53 – 33:17) 

I think it was published in the 80s. Yeah, it was. But it’s the same thing. I mean, you got, you got, you know, a manufacturing company that’s, you know, just not producing very well. And they can’t quite seem to figure out why. So anyway, so anybody who wants to a good read the goal, but it applies today still. I agree. A hundred percent. Yeah.

(33:18 – 33:55) 

That’s kind of funny that Jeremy said, you should read that book anyway. What advice do you have? And we just talked, we gave some advice. I mean, we are giving some advice, but what are some advice steps that you might have in addition to what we’ve already said? Yeah, so I think that the biggest one is about picking, picking your projects like we were just talking about there, that you want to approach that from first thinking about the business constraint and asking, how can we use these amazing new tools to attack that constraint? And then also a lot of times I’ll talk about a three-legged stool of strategy, implementation and follow through.

(33:56 – 34:10) 

So that’s the, that’s the strategy leg that is obvious, but is also often skipped. Right. And this is one of those situations where a lot of times having an outsider in the conversation helps a lot because you’re kind of too close to it.

(34:13 – 34:33) 

The saying here that I like is that your nose is an inch above your mouth and you still need somebody else to tell you that your breath stinks. That’s funny. That’s funny. So just talking through whatever this constraint is with somebody else can help a lot. Then the middle leg there is implementation. That’s the part I get all excited about.

(34:35 – 35:00) 

That’s actually like fingers on keyboard writing out the code in order to do this stuff. And then the third one also gets forgotten a lot, but I think is a big important part about what we do at Laconic Tech, which is follow through. So the first bit of that is making sure the team is trained up and is actually using this new tool because like a shockingly large number of corporate projects like this just wind up going unused rather than adopted.

(35:03 – 35:35) 

And then once it is getting used, we’ve got to start flowing real world data through the system and it almost never performs the way we thought it was going to perform in testing. Like 100% of the time, we wind up having to go in and adjust which data we’re giving the system access to, which tools it can call, the prompting, because we’re finding these edge cases in the real world that weren’t in our testing data set and we’re having to steer the system. Yeah, yeah.

(35:36 – 36:10) 

No, that’s really good. That’s a really good kind of overview from an expectation standpoint of how you should go about it. You know, if companies, I mean, who don’t, I mean, listen to some of the things like that and miss that opportunity to improve workflow, what do you think? What do you think about their ability to stay in the game for the next five or 10 years? You better have some other kind of moat.

(36:13 – 36:53) 

Right. I mean, there are like, it happens. Like we’ve talked about a couple of times that I come from the manufacturing world. Yeah. And like, to this day, you’ll have shops where like, it’s the, it’s the old man who started the shop, his wife runs the office and there’s a handful of guys working out in the shop and like, they take orders by fax machine, you know? So like, it’s, it’s possible to ignore modern technology if you have something else that’s keeping you in business, right? Maybe he does some kind of really pain in the butt manufacturing that nobody else wants to do, or he had a special set of relationships who has to go to the, to the house to fix it. Yeah.

(36:53 – 37:09) 

But it’s certainly not the case that most shops are taking their orders by fax machine in 2026. No. So like the, you better have some other kind of a moat that’s going to keep you afloat. If you’re going to put your head in the sand on this stuff, I think. I know. And can I, I’m older than I look to say, you know, my first job out of college.

(37:14 – 37:27) 

25? You’re 25? I’m 25. Yeah. The means of actually promoting, you know, what I had a newsletter that I worked for an organization. We mass faxed it. Email did not exist at the time. But you know, things have changed.

(37:28 – 37:43) 

Things have changed. It’s crazy. You know, I think back to your, your background in engineering, I think engineering still really the way engineers think and work on workflows lenses really well to actually implementing AI correctly.

(37:45 – 38:19) 

I mean, I think there’s a misnomer that you won’t need, but you need to have that process thinking of an engineer to make AI work for you. Absolutely. Yeah, I agree. It’s the, it’s the breaking a complex system down into simple parts so that you can then eat the elephant one bite at a time. Yes, yes. So I think it’s been really smart for you to, I mean, I know it’s been great for you and, but it’s been, it’s really, folks need to think through this, like, like the goal, like workflows.

(38:20 – 38:34) 

What is the problem you’re solving before you jump two feet in? But don’t ignore it because the elephant will stomp on you. You know, this is the Proof Podcast.

(38:35 – 39:28) 

What would you like to like leave folks with as they think through the next six months to a year and, and, and, you know, moving forward? Yeah, it’s, it’s probably a bit self-serving, but if you are in exactly the situation where you’re talking about where you realize there’s this wave coming and you can’t ignore it and you’d like to do something about it, but you want to do the smart thing instead of just saying, we’ve got to do more AI, I’d encourage you to reach out and I’d love to have that conversation with you. And of course I might send you off on a path that doesn’t even involve working with us and you can use some other tool, or if you wind up needing something custom built out, I know a guy. I know a guy, I know a guy named Ken who does, you know, helping, I mean, that story of the logistic company is just in crazy good.

(39:32 – 39:47) 

And it’s, it is an example after example that can be created if, if companies think more innovatively of how to leverage AI to make better decisions and, you know, win more business. And that’s exactly what they did. So Ken McLoud, thank you for joining us today.

(39:52 – 39:59) 

You’ve been on the Proof Point podcast and have a great day. Thank you very much. Thanks for listening to the Proof Point podcast. We’ll see you again next time and be sure to click subscribe to get future episodes.