Are AI agents really the future of software — or just the latest wave of hype?
In this episode of TechFirst, host John Koetsier sits down with Don Murray, CEO of Safe Software, to break down what’s actually happening with “agentic AI.” From AI-washing and “agent-washing” to real-world use cases in coding, automation, and enterprise software, we cut through as much noise as possible.
We explore how AI agents differ from traditional apps, why intent-based software is emerging, and how developers are already shipping faster with AI writing code. But it’s not all upside: there are real risks, from security vulnerabilities to the possibility of AI-driven mistakes at massive scale.
- Get the deepest insights concisely on the TechFirst Substack newsletter
- Subscribe to the TechFirst YouTube channel to never miss an episode
- Or listen on Spotify or Apple Podcasts
This episode of TechFirst is sponsored by Apprentice
Did you think AI was only for digital work? Nope … AI-native manufacturing is here. This month’s sponsor is Apprentice, which offers the first AI Agent built from the ground up for agentic manufacturing. Connects to all your systems, monitors everything, automates all your processes … but keeps a human in the loop. Check it out at apprentice.io.
Watch our conversation here.
You’ll hear:
- Why “agentic AI” is sometimes just a rebrand of automation … but not always!
- How AI is changing software development (and junior dev roles)
- The surprising productivity boost for senior engineers
- Why AI could make companies faster — and more fragile
- The rise of “good enough” content and the risk of mediocrity
- How enterprises are (and aren’t) keeping up
Plus: what happens when AI starts building itself — and whether we’re heading toward a breaking point.
Transcript: are agents the new apps?
Don Murray: We were at a trade show, and I said to the guy, “What do you guys do with AI?” And he was blatant. He said, “Nothing. We just put AI on our backdrop so that people would come by and talk.”
John Koetsier: Are agents the new apps? Hello, and welcome to TechFirst. My name is John Koetsier. Everything is agents, agents, agents these days, whether they’re open-claw-draining robot religions or narrow little agents focused on selling you something better with an email subject line. Are they the new apps, or is there something more profound happening with software?
We have Don Murray. He’s the CEO and founder of Safe Software. Welcome, Don. How are you doing?
Don Murray: I’m great. Thanks so much. It’s a pleasure to be here.
John Koetsier: The world’s gone nuts for agents. Why is that?
Don Murray: I think some of it’s hype, for sure. As I talk, I’ve been in the tech industry.
John Koetsier: Are you kidding me?
Don Murray: No, imagine. Never happens. No. Yeah, I know. Agents, apps, it doesn’t matter. At the end of the day, it’s software doing automation, right?
I think the strict definition of agents, the one that I kind of go with, is an agent will do something and make a decision, or at least do something and tell you, “Hey, here’s what I think you should do.” I’m not ready myself to have an agent go and… even a guy next to me tried to buy groceries with one.
It did a terrible job, and I told him we should try to buy a car. But I think it’s sort of like the new word for automation, right? I was at a show, and a guy said, “Don’t worry.” I said, “Tell me about your agents.” And he described his agents, and it was just a little bit of AI that hit a database and gave an answer.
I didn’t say anything, but I don’t think that’s an agent. I think that’s whitewashing the word “agent.”
John Koetsier: So we’ve had AI-washing for a couple years now, because the hype train has gone…
Don Murray: Off the charts.
John Koetsier: Now we’re having agent-washing.
Don Murray: Yeah, exactly. We were at a trade show, and I said to the guy, “What do you guys do with AI?” And he was blatant. He said, “Nothing. We just put AI on our backdrop so that people would come by and talk.” I really appreciated his honesty.
John Koetsier: Wow. That’s crazy. Yeah. Okay. So when I think agent, I’m thinking it has some level of autonomy. It has some level of intelligence. It’s got a task that is somewhat defined but maybe not fully defined. It’s applying some intelligence to that. It’s going off, doing the task, coming back, perhaps feeding its output into another agent, perhaps into a different system, or perhaps giving it back to you.
So that’s kind of my understanding of agents.
Don Murray: Yeah. Yeah. That’s great.
John Koetsier: Does that make sense to you?
Don Murray: Yep. And we certainly have built those. We used to call them automations. Emails are coming in, you’re scanning the emails, you’re looking for the intelligent ones. Now, I’d say with AI, the things you’re looking for… you can be much more intelligent about it because they understand context much better than the old way we used to search, where it was like word searching and looking for keywords. The action could be that it would send a Slack message to you or things like that, so you no longer have to watch for these critical things that come across.
John Koetsier: Did you think that AI was only for digital work? Not anymore. AI-native manufacturing is here. This month’s sponsor is Apprentice, which offers the first AI agent built from the ground up for agentic manufacturing. It connects all your systems, monitors everything, automates all your processes, but keeps a human in the loop.
Check it out at apprentice.io.
That’s super interesting. I just had Andy Marcus, the chief data officer of AT&T, on, and he was talking about how they’re using agents right now to diagnose network outages or challenges, and it’s hundreds of times faster than a human doing it, which makes sense. Okay, so I started off with a stupid question, right? Agents, are they the new apps and stuff like that? But fundamentally, this does kind of represent a new understanding, a new architecture of software, correct?
Don Murray: And we’re seeing that. One of the things we’re building now is just users entering their intent, what they want the software to do, and then AI with agents is figuring out exactly how it’s going to make that happen. Whereas in the past, it was a lot of developers sitting down and building these deterministic steps, step by step by step by step.
Whereas now, anybody who just has a good question and understands the domain can write the intent and then build the actions to do it. Now it’s really interesting because every time we feed the same intent into our system, AI is spitting out a different answer.
John Koetsier: Mm-hmm.
Don Murray: Now, when we run them all, they all give the right answer, but it’s a bit interesting that AI is finding things and we’re like, “Wow, does that actually work?”
So we code it up just to verify, and yeah, that does work. But clearly, sometimes you can use that to get the thing, and then you want to nail it down because you don’t really want a system every single time doing all that work. Sometimes the work of building that is worth more than the action itself. If you can see where I’m going.
John Koetsier: It really depends on what it is, right? If it’s a high-volume action, using an agent, using any kind of LLM or AI, is much heavier computationally than the old-fashioned “if this, then do this” programming and everything. However, there are cases where you want to use it.
I’ll give you an example. I’ve recently started building apps again. I did that in the past, haven’t done it for a while, so I had to reopen my App Store account. Guess what? Now I need to enter a bunch of information, some tax information. I need a DUNS number. I didn’t need a DUNS number before.
So then I’m wondering, do I need an agent? Can I just give it all my information and say, “Go do it”? And there isn’t one. And that’s kind of one of the questions I wanted to ask you: can the old, big, dumb companies, and I’m naming Apple here, it’s a tech company, one of the largest companies in the world, it’s super high-tech, but their systems are pretty antiquated. Can these big old dumb companies survive this transition, or are we going to see others just come in and eat their lunch?
Don Murray: Yeah, and for sure some won’t survive. The tech world is just full of companies who’ve washed up on the beach after wave after wave, right?
Going way back, and I’m showing my age, was DEC. I mean, I read about them, where they introduced a minicomputer and really ate some of IBM’s lunch on mainframes, but then they didn’t believe, “Oh no, PCs, they’re not going to do anything.” And so they washed up on the beach.
The question is, Apple in 10 or 15 years… these big ships, they take a long time to go down, right? So are they still the company? It’s a good question. We talk about it at Safe. We talk about day-one companies versus day-two companies.
John Koetsier: That’s Amazon.
Don Murray: Yeah, it is. We don’t want to be a day-two company that gets too comfortable with who we are and thinks we’ve got it all figured out, or worries that, “Oh, this new thing is going to sacrifice our old thing.” We always say we’d better sacrifice the old thing, because if we don’t, somebody else will.
I think it’s really a mindset. Look what happened to Intel, right? They were at the top of their game, and now they’re still around, but will they come back? Time will tell.
John Koetsier: Time will tell. Exactly. Yeah, it is interesting. I mean, with Apple, they have this massive moat of owning the mobile phone, right? Of the wealthiest chunk of humanity, essentially. So you’re going to go through whatever pain you have to go through to cross that moat and get inside and put your stuff there.
But let’s talk about making software agentic. What’s the benefit here? What’s the core goal here?
Don Murray: Yeah, so certainly we use it to write code. We use it to help us write code, and what’s really interesting is we use a copilot. Now developers, again, can talk about what they want to do, and then the AI system gives them back code.
What’s really interesting is the developers with a lot of experience get a much higher productivity improvement than a developer who’s new, because they have so much experience that they can just tell, “It’s hallucinating there. Oh, this is useful. This isn’t useful.” Whereas somebody junior can’t really make that distinction as quickly.
So I had a senior developer who’s a superstar, and we’re doing this MCP thing, and he said, “This Copilot saved me weeks of work.” And I’m thinking to myself, saving this guy weeks of work? This guy’s super fast. So that’s what we’re really, really interested in seeing.
We also have things that are more agentic, constantly scanning our code, looking for vulnerabilities. Because we know there are agents out there trying to crack code now, and there are agents out there looking at your code to try to identify vulnerabilities. So you sort of have this agent war, I’ll call it, with some agents trying to break you and some agents trying to protect you.
John Koetsier: The world is… it’s the latest escalation.
Don Murray: But still…
John Koetsier: It is interesting. Go ahead. It is interesting, actually, talking about developing code, because I’ve seen developers who are complaining, and software development managers who are complaining, that their junior devs can’t explain their code.
Don Murray: Yeah.
John Koetsier: Because they didn’t write their code.
Don Murray: And see, we have this rule that every line… if you use AI, more power to you, but you have to code review that and make sure you understand it. Then we have a second code reviewer who must understand it, because the cost of code is the maintenance.
We’ve had genius developers in, but they wrote such compact code that nobody could understand it, right? Because he used all the fancy things in the language, and it was buggy too. But when it was buggy, nobody could figure out what was going on. So we learned to unwrap it and just have simple things that anybody can understand.
John Koetsier: From the consumer perspective, however, what’s the benefit of agentic software, software that has agentic features in it?
Don Murray: Yeah, and I think it can just, again, save you time. If I think about this grocery thing, I could dig up the name of the tool. He’s plugged it in. He doesn’t run it on a company computer because it literally sees every single thing on his computer.
John Koetsier: So…
Don Murray: It’s…
John Koetsier: Open Claw.
Don Murray: That’s what it is. It’s Open Claw. Yeah, that’s the one. And you set it up to do things in the back using all his data. That’s the one he used to try to set up groceries.
He was convinced it was going to work. But was it a user problem or was it an Open Claw problem? Fortunately, he ordered too few groceries and not too many.
John Koetsier: So essentially, eventually he could connect that to a smart fridge, just taking a picture of what’s in the fridge or something.
Don Murray: Yeah, exactly. And then it would just… the idea is, I mean, what do computers do? They really are just here to save us all time, right? Get new insights.
John Koetsier: Is that what they do, really?
Don Murray: I think so. For me, my whole career is building these things. Yeah, that’s awesome. I love it.
John Koetsier: There was just a Harvard Business Review study that came out, actually, and it looked at a huge swath of companies and people and the impact of AI, and it turns out that AI actually makes us busier.
Because we can do more. So you’ve got QA people who are writing code, right? You’ve got other people who are expanding what we’re doing because AI is helping us do it, and that means we’re looking at it more. We’re a little more stressed there.
Don Murray: Yeah, and it’s exhausting, right? Working with AI, I work with AI every day, and it used to be you’d have a junior co-op student or something, and you’d give them a task and you knew they would come back hours later, right?
John Koetsier: Yeah.
Don Murray: Now it comes back in seconds, and you’re like, “Oh, okay.” So then you don’t even have time to go get a drink or get a coffee, and then you go again and you go again.
John Koetsier: I find it hard to stay in flow when I work with AI because of that. It’s not an hour’s wait. It’s not a two-second wait. It’s a 15-second or 30- or 45-second wait. So what do I do? Stare here at the screen while it’s doing this? Do something else? So it’s hard to stay in flow when I do that. Yeah, that’s really interesting, actually.
Don Murray: That is. That is, yeah. And it’s too complimentary. Not all my questions are good questions. “Oh, great question.” I want AI to tell me, “That was a really dumb question.”
John Koetsier: You can probably give it some context and say… exactly, exactly. Now what’s a downside? Let’s say you’re an enterprise software company and you make your software agentic so that somebody can say, “Make me this,” instead of actually doing it.
So you start to see that in some Adobe software, right? Where they give you a prompt box, and you tell the software what you want rather than having to click around in boxes and menus and add things and everything like that. What’s the downside? Does it become more hackable? I mean, do you worry about prompt injection for those things?
Don Murray: Security definitely is a thing, but I think you lose some creativity, right? Because if you’re doing it in the old way, you’re drawing and you’re doing, then I think you’re engaging your mind more. It’s truly going to be a unique creation.
Whereas I can’t draw at all, so I think it’s great. But you can tell when things are like, “Oh no, not another AI graphic.” I find it good when I’m working with marketing to use it to illustrate what I want, just as a way to communicate to them. Then they take it and add the creative flair.
But I think, yeah, you can do a lot of cool things, and they’re not necessarily… I mean, I like cat videos like the rest of them, but AI can do a good job, and you just wonder how much value we’re getting and how much time we’re just spending wasting. But I think we’re losing that creativity because we’re kind of working in the constraints of AI, whatever that tool is capable of.
John Koetsier: The way I put it is that it has never been easier to be mediocre.
Don Murray: Yeah.
John Koetsier: So you can absolutely achieve some incredible results with AI. Don’t get me wrong, and we’ll get into that. We’ll talk about what we’re seeing with Anthropic and with OpenAI. We’ll talk about that in a moment as well. But the default sort of standard user of AI going to ChatGPT and doing something, it’s easy to be mediocre.
And so essentially, it takes the bar and raises it for everybody. Okay, the bar was down here. It was way, way, way low. Now the bar is here, and the bar is like C-plus, right? And sometimes for some things C-plus is just fine. It’s okay to have a C-plus in something. But if you really want world class, if you want absolute world class, you either have to work much harder with the AI, or you need to be the type of person who can produce A-plus output.
Don Murray: Yeah, yeah. No, I think that’s… I really like that. What we say here is that, remember, AI is your assistant. When it gives you something, you have to take it and take it to the next level, right?
I’ll use it to help me with titles for abstracts, right? “Oh, I’ve got a presentation. Okay, give me five titles.” I never actually take one of the five, but certainly one of the five will inspire me. Then I’ll just give it a whole bunch of points and say, “Write me an abstract that’s 450 characters,” and it’ll give me that. Then I have to go through and edit it. But it gives me some sentence structure and some flow because it understands the structure of what a presentation is.
John Koetsier: Yeah, a lot of people agree with you, and they say that AI solves the blank-page problem.
Don Murray: Yeah.
John Koetsier: Getting started. A lot of people have this blank-page problem where, where do I start? What do I do? How do I…? Exactly. And it gives you a starting point, which you can totally reject, and often should, but at least it gives you something to react to.
Don Murray: Yeah. I never viewed it that way, but you’re absolutely right, because sometimes it’s like, oh, what’s the hardest part? It’s getting started, right?
John Koetsier: Yeah, yeah, yeah. Exactly. Exactly.
Don Murray: Yeah. Yeah.
John Koetsier: Others are saying that AI is removing all excuses because everybody has all these 15,000 different ideas. I could build this business, I could build that business. Now you can, right?
Don Murray: Yeah. Theoretically.
John Koetsier: Can.
Don Murray: Well, I think “can” is interesting. We had sales guys come by and they said, “Oh, we found this AI tool that we want to use.” And it was like 80 grand a year. It wasn’t going to bust the company, but that’s some significant cash.
So I said, “Jeez, just try. See if you can build it in Gemini yourself.” And this is a sales guy. They came back a day later and they go, “Yeah, we just used Gemini and some quick query prompts, and now it’s 90% of what that tool was.” And we’re on Google, so again, that company…
I was also at an investor kind of show. It was just interesting. We’re not looking for money, as my whiteboard shows you, but they had a company up there and they said whenever they’re looking to buy a company, what they do is hire four really smart people and give them a month and see how much of that solution they can reproduce.
If they can reproduce 80 to 90% of it, then they decide, “No, there’s not really anything there,” which is interesting.
John Koetsier: That is interesting, especially since every single software project I’ve ever been in, you think you’re done, you think you’re 90% there, and you find out, shoot, I’m only 50%.
Don Murray: Yeah, that’s true. Yeah. It’s like, “Oh yeah, what are you going to do in two weeks?” Then you come back in two weeks and, “Weren’t you going to do that in two weeks?”
John Koetsier: Right, exactly. It is this magical, receding deadline.
Don Murray: Yeah, exactly.
John Koetsier: I do want to talk about AI making AI, and so a couple things there, right? We saw the latest version of code for Anthropic was written almost entirely by AI. We had the CEO saying that I barely write any code anymore. AI writes all my code. It doesn’t mean he doesn’t check. It doesn’t mean he doesn’t look at it. It doesn’t mean edit. It doesn’t mean all the other things there.
And also, it doesn’t necessarily mean that you use what you get on your first try. Often, to get the best output, you’ve got to fine-tune your prompts and all those sorts of things.
So OpenAI, they recently disclosed in their notes as well that the latest version of ChatGPT was also built partially by AI. So some of the most sophisticated code in the world that we’re using for everything is building itself, which kind of says something about the future of software development, I would think.
Don Murray: It’s interesting, isn’t it? There are even social networks that are only for AI.
John Koetsier: Yeah, that’s Maltbook. I wrote about it a week and a half ago.
Don Murray: Yeah. They’re developing… AI is developing its own language, and they’ll punch you off if they realize you’re a sentient being. It’s really, really fascinating, I would say.
What they’re talking about… I mean, even at Safe, we do have… we’re using Claude. It’s a big one that the developers are really excited about. We have this rule: you can use whatever AI you want. It has to pass, obviously, some security and legal things. But initially we used OpenAI, now then to Gemini. I only use Gemini and Claude. I don’t even use OpenAI anymore.
And the coders love Claude. So more and more of the code is being written by Claude, but again, they’re inspecting it. They’re having to sign it off.
John Koetsier: Mm-hmm.
Don Murray: Then it gets code reviewed again, just like it would if a human wrote it. So a human writes it, then another human checks it. AI writes it, then you have two humans checking it.
John Koetsier: I like that. I like that. But it’s insane what this is doing to our world in terms of speed-up. You talked about that superstar you had who said it saved him weeks in doing this stuff. So you’re shipping features faster and faster and faster.
I wrote this in the piece that I wrote about Maltbook and about Open Claw. I said it feels like the singularity, or approaching the singularity, when that pace of technological progress just gets so fast, exponential. It’s already been hard to keep up with tech, what’s new and what’s going on.
That’s also a problem if you’re a SaaS company or you sell software. I mean, how will your customers keep up with the flood of new features that you can do quicker and quicker and quicker and quicker?
Don Murray: I don’t know. I just don’t know. We are delivering stuff faster than ever, and our customers already have to plan when they upgrade because they can’t just slap in a new version. They want to make sure all their workflows still work.
And now we’re saying, “Oh, the way you did that before, now you can do it this way. It’s faster, it’s less effort.” And many of them just want to know that it’s there. They don’t even plan on using it for a while because they’re like, “Well, let’s wait six months and let’s see what’s there in six months.”
John Koetsier: Well, yeah, then it’ll be different.
Don Murray: Yeah.
John Koetsier: But users hate that, right? They know what they’re doing. They know their routine. This is how I use this software, and now you’re moving their cheese.
Don Murray: Yeah. And if they do the upgrade now, by the time they finish the upgrade, they’re just going to be further behind, right?
John Koetsier: It’s a real problem.
Don Murray: Yeah. But the one thing that is key is what we do is data, right? So we’ve been lucky there. But still, now we have synthetic data because we ran out of real data. So now we’re making synthetic data.
John Koetsier: That is true. And I guess the solution for this software or technological lag that we’re talking about here, and also just this, “Oh man, I can’t keep up,” is if you have an agent who’s helping you, you just say what you want. It couldn’t be simpler, right? This is what I want to do, and the agent should help you.
So essentially every software…
Don Murray: Yeah.
John Koetsier: Should ship with an agent.
Don Murray: Yeah, absolutely.
John Koetsier: You might want to turn agent mode off because you’re an expert and you do this or whatever, but you know what? You should have an agent mode in anything, whether that’s Salesforce, whether that’s an app that I’m using on my phone. Agent mode is just, “I want to do X. How do I do X?” And the agent helps you.
Don Murray: Even on databases. Why learn SQL? You can just talk to your database and it’ll just figure it out.
John Koetsier: I can have 10 indices on the same table.
Don Murray: Yeah. Yeah, I know. And if you’re a domain expert who really understands your domain and what you want to do, this is a boon for you, right? Because in the past, if you didn’t want to learn SQL… not everybody in the world wants to learn SQL.
So in the past, you’d have to try to talk to somebody who is a SQL expert, try to take what you wanted, turn it into SQL, and then turn it into a database query. Now you can just talk to it. You’re going to get your responses quicker, and then you can do, “Oh, what if? What about this? What about this? What about…?” And you can just have that.
When things become so quick, it just changes the dynamic. If you ask me something and I come back to you in a day, then okay. But if it comes back to you in seconds, now your mind’s still there, so you can have those conversations.
John Koetsier: Yeah, I can’t stop thinking about the Spider-Man quote: with great power comes great responsibility. And I can’t help but think that in all this world of acceleration and better and faster and more, whether that’s software-related or process-related, whatever, we’re going to have people who are going to connect tools to their code base, to their database, to their entire setup, who are going to massively screw everything up.
Because they’re going to say something wrong and they’re going to do something wrong, and they might obliterate a billion-dollar company in 0.2 seconds.
Don Murray: Oh yeah. Yeah. Tools like Open Claw, you just literally give it your whole machine. I don’t even know what’s on my machine, and it’ll just figure it out. If you unleash that across an organization, there’s data there that nobody knows is there.
And AI isn’t perfect. It can’t tell what data’s real, what data’s… it can’t tell fact from fiction. And it could just do the wrong thing with the wrong data and, you’re right, obliterate a company in no time. Or open a floodgate on a dam. There are just lots of things that can start off badly.
John Koetsier: Let’s start the countdown right now, because we’re putting AI into companies and software faster than ever before. How long will it be before a company itself… this drops? Wow.
Don Murray: I don’t know. It’s interesting. I have staff who say, “Do you think there’s an AI bubble?” And I’m like, “Probably.” But there was an internet bubble too. But the internet’s still here. If there is a financial bubble and it breaks, it’s not like with AI we go back to the way we were before AI.
John Koetsier: Mm-hmm.
Don Murray: AI is here.
John Koetsier: I tend to agree with that. I mean, there’s obviously a huge amount of investment, GPUs all over the place, alternative architectures for chips and everything like that, massive build-out of data centers all over the place, power demands that the U.S. grid probably can’t handle. China’s doing well because a lot of green power has come on.
So there is the bubble and there is the overhype, but the dot-com bubble burst. Pets.com went away, but the internet didn’t, as you said. And there was a lot of fiber. It became a trillion-dollar thing.
Don Murray: Yeah. There’s a lot of fiber in the ground that wasn’t being used then, but it’s probably being used now, right?
John Koetsier: Exactly. Yeah, exactly. Yeah.
Don Murray: And I think already we’re seeing it. Some of the big vendors missed their targets, and I was reading an article, and what they were saying is they’re having trouble finding power, right?
John Koetsier: That’s why Microsoft is building nuclear reactors.
Don Murray: Yeah, yeah, yeah, yeah, yeah. It’s just crazy. And these data centers are absolutely monstrous, right? I was talking to a guy in Alabama… oh no, Louisiana. They’re building this massive data center, and he says it’s just the biggest thing he’s ever seen, and it’s causing water problems.
Because for every watt of power, you need a watt of cooling.
John Koetsier: Mm-hmm.
Don Murray: So it’s… yeah. Really, yeah.
John Koetsier: It’s got its challenges. There’s no doubt. Now somebody did analysis of data centers versus an In-N-Out Burger. It turns out an In-N-Out Burger consumes more water than many data centers because it takes a lot of water to make beef.
Don Murray: It does. Never thought of that.
John Koetsier: So…
Don Murray: Yeah, it does. It does. Yeah.
John Koetsier: There are multiple things going on there. Yeah. I mean, of course, you’ve got Elon Musk…
Don Murray: Yeah.
John Koetsier: Go ahead.
Don Murray: This is where it’s going to do amazingly good things too, right? When you think about healthcare, the ability to really analyze images and detect disease and things.
I think healthcare and AI… you just think about the vast knowledge there is in health, and there’s no way any one doctor is going to be able to…
John Koetsier: Be an expert in anything. Exactly. We’re just at the cusp of that right now, actually. I just had a press release cross my desk a couple days ago. A major HMO in the United States, and they’re adopting healthcare AI across their practice, and it’s increasing diagnosis accuracy. It’s increasing sharing of information.
Don Murray: Yeah.
John Koetsier: I mean, if you’re a doctor and you’re going nuts, there’s too little healthcare in the world. You talk about the developing world specifically, especially. So if this can help, that’s a wonderful thing.
Don Murray: Yeah, it is. And we use it. One that we’ve had a lot of success with is we get, of course, a lot of legal documents, right? And they take a lot of time. Having an AI tool that can quickly read them and is only trained in BC law or Canadian law, right… you can’t use one on the open net because you have no idea what law it’s working with.
So it’s really enabled, again, as an assistant, really helped our legal staff be more efficient and get more stuff done.
John Koetsier: Mm-hmm.
Don Murray: I think anything where you’ve just got this massive, vast, huge body of knowledge that you’re supposed to be basing your decisions on, AI really helps with.
John Koetsier: Absolutely. Absolutely. Cool. This has been a great conversation. We’ve gone a lot of different places.
Don Murray: Yeah.
John Koetsier: I don’t know where things are all going in terms of AI and agents, but I assume bigger and better and faster are among those, and maybe in some cases, as we talked about, way worse for certain companies that adopt it in ways that are…
Don Murray: Yeah, yeah.
John Koetsier: Unwise.
Don Murray: Yeah. Agentic AI… I was on two calls today, and agentic AI came up in both of them, right?
John Koetsier: Mm-hmm. Mm-hmm.
Don Murray: And so it definitely is the other big word that we’re hearing. The other big word that we’re hearing is “AI-ready,” because we’re in data. So, “Is your data AI-ready? AI-ready?” And it’s about data quality. It’s not just, “Is your data accurate?” but “Is your data biased?” which is a whole new type of data quality, right?
John Koetsier: Yeah.
Don Murray: Really interesting.
John Koetsier: Yeah. Cool. Well, thank you for your time. Hope you have a wonderful afternoon.
Don Murray: Yeah, thank you. Really appreciate it. Yeah, take care.