Foreign welcome to Emerging excellence and it's our pleasure today to welcome Sam Malakargalan.

And Sam, you're coming in from the beautiful usa.

You're the general manager at Agent AI, which I think typifies some of the most interesting and kind of, kind of perhaps paints a really interesting future picture of what technology is coming down the pipeline and so opening up some of the opportunities.

So first of all, Sam, a huge welcome and thanks so much for being on the show.

Thank you for having me.

I'm looking forward to it.

So, Sam, I think for most people agent AI would be a completely foreign concept.

I certainly am no whiz in this space.

And I think if you could break it down for us for a second.

What, what is agent AI and what is an AI agent more broadly?

Yeah, it's a great example of why we should never let engineers name things.

ChatGPT is a meaningless thing.

Calling it an AI agent is a technical distinction.

AI we're all kind of familiar with it's, you know, lead scoring.

We can use it to predict the weather.

The difference in AI that you're used to, you know, maybe it's generating ads or whatever and agents is that agents have gotten so good at predicting what's going to come next that we started trusting them to just do it.

So I can say, respond to this email for me.

You've seen autocomplete same general concept, but instead of auto completing a sentence, you can autocomplete an essay or a book or a blog article or an entire email.

So an AI agent is just a piece of software that's using what's called a large language model, usually to try and predict how you would respond given as much kind of context as you can give it.

And then we call it an agent because it has agency.

Agency not in the philosophical sense, but in like the literal sense.

Like it can actually go do stuff on your behalf like a buying agent or a human would.

We now have these armies of little AI agents that go do stuff for us all the time.

So you've got a very interesting.

So your background, if we just touch on that briefly.

So you're ex HubSpot and you were kind of super early in that game, in that game that's gone on to become, you know, obviously a massive company these days in terms of like where did.

What.

What was in your background or your kind of story which had you focus on this piece.

What, and particularly why are you so.

Why, why are you so kind of excited about the potential here?

So I really like unfair fights.

I like creating unfair advantages.

And this is just such a great, I mean, you think about Google got sucker punched by a NonProfit Research Company.

OpenAI was originally a nonprofit research company.

ChatGPT was just a demo.

Uh, and if Google can get sucker punched, what does that mean for like the 800 pound gorilla in Everybody else's industry?

Uh, but what worried me was I, I worried that we would go from having millions of websites to having like thousands of agents and just dozens of, you know, kind of orchestration agencies, enterprise silos.

That would be expensive, you need technical skill, the scope would be limited.

Uh, and so when Dharmesh had the idea for agent AI, it's like, what if we could make it so that anybody, regardless of technical skill, what tools you're using or your ability to afford, you know, AI tools, what if everybody could be a part of kind of the AI future?

So not to sound too melodramatic, but we're trying to avoid the dystopian hellhole that everybody's worried about us ending up in by making it so that everybody has access to the same tools.

And the only thing limiting use your creativity, not your budget.

Such a cool idea.

And I think, I think we can definitely can see that happening.

Right?

Especially as they talk about how there's going to be those who understand AI and those who don't, and it's just going to be basically a free for all for those who do.

And I think it's very cool.

Some of the examples we just touched on just before we jumped on in terms of what people are using technology for.

So just talk us through.

Obviously we've got our construction and infrastructure bent in terms of.

That's where we kind of focus right now.

Tell us from your perspective, what are some of the interesting applications that you've seen people develop in the space can actually use agent AI for at the moment?

Yeah, I love people who use it for communication.

One of my favorite stories, I had a friend, if you don't tell AI to use a certain tone, you can kind of tell it's AI.

So I had a friend who, a couple weeks ago, I was asking, are you writing, using AI to write your text messages to me?

And he was, he's got some, you know, learning communication disabilities and he was using AI to turn his, like, long rants into the types of content that his friends and family would, would want to see.

I love people who use it for learning.

So the other night I couldn't sleep and I spent four hours having AI teach me some Weird nuance about quantum mechanics and it no human one.

I couldn't afford to have a human teacher in my room at 4am, nor would I want to.

Also, no human teacher would have that kind of patience.

But it sat there, it's like, eventually it's like, oh, I see why you're not getting this.

So using it to learn new skills.

Wow.

And then using it to give you skills like AI is very good at being mediocre at things.

So it's a mediocre software engineer, it's a mediocre marketing professional, mediocre designer.

But what that means is that all of us have access to mediocre skills.

So like I was telling you about the, the, my local Irish pub, they wanted to have like a drunken spelling bee game, which is cool, right?

Just something different than trivia.

And normally I would have had to go bother my engineering friends to help me like set everything up, but I just used AI.

I said, this is what I want it to do.

Generate all the words, write all the code, deploy all the code.

And so they have, you know, I think it's drunkenspellingbee.com you can probably check it out.

It's entirely powered by AI and it's like we're all still T shaped.

We have our area of expertise, but the width has gotten much wider.

Sales reps can generate decks without talking to the marketing team.

Individual project managers can do data analytics without having to bother anybody.

I find that to be the most interesting thing, is that every individual person has access to a far greater number of skills than they used to have without having to actually learn at all.

Yeah.

And I think it's interesting, like from my side, I'm, I'm dyslexic and my spelling is average at best.

But now, you know, I can run any post like LinkedIn or something through, you know, check for spelling, check for words, like, and it can just bring it up to a level which is significantly better than what it was before.

And I think that to your point, right, the base level, I was talking to someone just the other day and she works in PR and she's like, I'm really worried about AI taking my job in pr and I'm like, probably will like it.

Probably take all of our, you know, this at this base level, right?

Because then if you don't rise to that level, there is an opportunity that those who, who leverage it actually could become supreme in many ways.

Would you, is that, is that a fair assessment, Sam?

Yeah, it's absolutely fair.

It was one of My biggest fears with AI was, for example, if you replace all the entry level sales reps to do chat on a website or whatever, where do the senior sales reps of tomorrow come from?

What's interesting to me is now people are building AIs to help people practice those.

So instead of trying to replace a sales rep, replace a prospect.

Let the sales rep practice against a fake prospect and learn the skills that they need to have.

Wow.

So I.

It's definitely something where if you are early to it and if you can deal with the fact that it's confusing, I mean, chaos is an opportunity, right?

And this is chaos right now.

This is the most chaotic thing I've ever even studied, much less actually experienced because it's so fast, right.

The Industrial revolution took a century, didn't happen everywhere at once.

Even digital transformation took like 20 years before not adopting, not having a website put you at like a huge competitive disadvantage.

You have like 20 months until not having some level of familiarity with AI puts you at a competitive disadvantage in your career as a company.

The good news is everybody else is confused too.

Like nobody really knows what's going to happen next.

We all have suspicions, but much like the early days of the Internet, the App Store, like when Apple rolled out the App Store, they didn't anticipate Uber, right?

But it was, they gave a platform for that innovation and then, you know, the Uber folks made bajillions of dollars and that's what's going to happen the next 5, 5 ish years.

Tell us about the commercial model behind Agent AI.

So let's say me as a, you know, just someone sat in Australia, I can jump on to, what can I, what can I do?

What does it cost to kind of set up a little agent for myself or, or in my company?

What's that look like?

Yeah, so it doesn't cost anything.

And the reason it doesn't cost anything is because the costs to us are going down dramatically over time.

So the performance of AI models is increasing really fast.

The cost of AI models is decreasing really fast.

And it would be kind of like if Squarespace or a website provider was still charging you per website visit, even though cloud hosting is ridiculously cheap.

That would just be a really bad business model.

And then the bigger thing for us is every, we've got all these people, we've got, I think 3,000, 4,000 people on the site who are building agents and over 1.1 million people who are using agents.

And eventually you're going to want to just like the App Store be able to allow people to make money off of it.

The benefit to us is that AI gets smarter the more data is run through it.

So the more people create and use AI agents, we're going to be the best place to help you.

You know, our AI's job is to know you really well in your job, your company, your competitors, your style.

And then based on what you're trying to do, we're going to tell you use this agent, give it that information, then use that agent's output to do something else.

Like our AI agent is learning from everybody who's using it.

And I don't care if you're Google, Microsoft or anybody else.

Like no amount of money can buy time, can buy observations of humans doing things.

And so that's the commercial model for us is can we, can we.

And can we also just make it like take the wind out of the sales for these enterprise silos?

You know, love my dear friends at Microsoft and Salesforce and such, but I don't want to see a world in which AI is only useful in the context of specific tools where you have to have technical skills to use it, or more importantly where you have to have a lot of budget to use it.

Right.

Like I want my.

I kid you not.

My 16 year old nephew living in rural Tennessee here in the US is one of the most prolific AI agent builders on our site.

And he doesn't have budget or anything else like that, but he's able to be competitive with anybody he wants to try and be competitive with.

Those are the stories I love.

Level, let's say I jump on, I start to create my own thing.

Can I sell that?

Is it possible to sell that as functionality to a client or something like that?

We will eventually support that.

Keeping in mind that we just launched last May.

Technically we launched in last September at HubSpot's inbound conference.

But that is the idea is we want to make it so that you can either make it public, you can keep it private.

If you want to just use it for yourself, that's fine, you can make it public if you want to just do that.

Or you can charge a monthly fee, you can charge just a lifetime license just like you would with any app store.

There's all these companies that make tons of money off the app store and that's what we really want to get to.

It's just quite a complex problem to solve.

So right now it's one of the reasons we're about actually by the time this comes out we'll have Announced it.

We're launching a $1 million grant fund to subsidize builders building agents simply because we can't yet figure out how to do the monetization between, like, somebody uses your agent, you actually get paid.

So we're just giving people grants to build really cool agents.

And we know that the enterprise value to us will come, will follow that.

Interesting.

Very cool.

So let's talk about construction.

There's a whole.

Construction is a very tricky space.

There's a lot of moving parts.

There's, you know, heap of scheduling and in the, in the actual kind of building process.

But then there's a whole bunch of communication going on between designers.

And so they're getting information when they're digging holes, as an example, and they're, you know, discovering pipes which were never there.

And so they have to kind of adjust designs.

There's a whole bunch of procurement which is happening, you know, even before they kind of try to find the gear.

There's this small markets like tunneling companies selling dirt to the road building because, you know, like, they're trying to get rid of it, the tunnel, but they, they need dirt when they.

To go up the hill in the road project.

What, what do you think of the opportunities within that space or kind of, what are the practical examples where, as you said, you can, the project manager is.

No, doesn't have to bother the, the data management team or the, or the data analysts anymore.

They can kind of just tap into something and have the information.

What do you see just in terms of what you know about construction?

You certainly don't have to be an expert here, Sam.

But what do you think is kind of the obvious examples and maybe what's coming beyond that?

There's two big buckets.

One is, can you take things that you're already doing and make them more efficient, make them faster, make it so you don't have to do it right?

Like, I have AI that attends meetings for me and just lets me know if there was anything that it thinks I would find interesting on that meeting rather than me having to join all of them.

There's, you know, and I was having a conversation with somebody the other day.

They're like, so is it just going to be my AI agent and your AI agent having a meeting?

I'm like, is that bad?

Is that like something we don't want as long as it has enough context around us?

Same thing for buying and selling, right?

Like, is it going to be my AI agent whose job is to buy things I will like and your Your agent's job to sell and they'll do all the nonsense, they'll do the demo call and you know, looking at the procurement, all the like, the details that take us tons of time.

So anytime you can think about what you're doing now, is there a way to use the AI and use specifically these AI agents to help you do it better, help you not have to do it at all and help you do it at greater scale?

The other thing is just net new capabilities.

So can you like, like you said, you know, sales reps can design, project managers can do data analysis even, you know, AI has this fascinating feature or bug, depending on your perspective, called hallucination, where it will make up things that don't actually exist.

I find that fascinating because as long as you're aware of it and you don't use AI to write your legal brief and then get made fun of by the judge because you cited a case that doesn't exist, which actually did happen all the time.

Yeah.

If the AI is hallucinating a product that your customers might want and you don't sell it, maybe you should.

Or if it's hallucinating, you know, different types of companies that you could be selling the excess dirt to and you haven't thought about it, like that's a feature.

So giving yourself so one, just make everything more efficient.

Like we, you shouldn't do all the things that just take tons of time to give yourself those net new capabilities.

Every person on your team should be a thousand times more powerful than they were a few years ago.

But then three, you know, let AI be co creative with you.

I sometimes tell it to hallucinate intentionally, just to see what it comes up with, either positioning or products or something else that we should do.

So for the construction industry, obviously efficiency matters a lot.

But then very importantly is going to be how do you manage all those processes?

How do you make it so that it knows if it's going to rain during the project and automatically reschedules everybody and sends a note to everybody?

Or how does it know that you know, a bridge collapsed and that means that you know something that you need isn't going to make it there in time, you need to reschedule things.

All of this kind of information synthesis is incredibly powerful for anybody.

But in particular, when you've got all these complex moving parts and you've got all these people you have to talk to, AI is going to be beautiful.

For that, I feel like, especially if you're working in a, in a global world and you when it comes to like structural building up with your trading metal or even currency, if you're buying in US dollars but your domestic currency's Australian dollars as an example, then understanding some global shifts and making predictors where typically that would be the realms of investment banks perhaps, or you're having to leverage expertise elsewhere which has become the domain of these huge advisory firms.

What you're saying is that the local baker who's buying flour or something like is going to have access to very similar information potentially from, from that regard.

Is that, is that right?

Yeah, I find that fascinating if you, you know, one of the first things I did was I built a McKinsey consultant in a box as a joke when, you know, this is a year and a half ago, but last summer I gave a demo to the executives at a very large bank and I just asked, I had an AI agent, very simple question.

I said the price of oil went up, should I buy or sell the like the Canadian dollar yen pair?

And it's told me what to do, gave the expl.

And one of the people in the room was like, damn.

Like that's my job is to prepare reports for people and that person will figure out something else to do with their life.

I understand that there's that kind of job displacement.

But what's more interesting to me is now Everybody has a McKinsey consultant.

Everybody has like a logistics consultant in their pocket that can, you know, pay attention to all things that you can like even for marketers, right?

It's like, you know, the Federal Reserve kept interest, raised interest rates.

Here's what that means for your pay per click campaigns.

Like all of this stuff that we have to couldn't possibly synthesize on our own unless you had a big team or big budgets or access to consultants.

Now everybody has that from a leadership perspective.

So we train first time leaders, emerging leaders, new leaders and increasingly just even just any existing leader.

Because we found that a lot of leaders don't have any understanding about what they really should be doing broadly, which is despite their seniority, some of their foundations can be somewhat shaky.

From that perspective, I see as we look forward that leaders are not just going to be leading people directly now, going to be leading technology.

If you work for Amazon in a distribution sensor, you've already probably got a group of robots which are currently in your workforce.

I would imagine that's going to be more prolific in terms of, as we move forward, especially when it comes to construction safety and things like that.

And also labor here and domestically is extremely Expensive.

And so if you can.

And also we can't find any more people.

That's the other part.

It's not just expensive.

It's like there's no.

There's this huge surplus of.

Of demand and insufficient people to do a bunch of the work.

So if you can displace some of that or replace it with a robot to do, you know, moving the cement bags just as a real kind of trivial example, then all of a sudden you free someone up to do potentially something else.

So what do you see?

What's a leader going to need to be prepared for when it comes to the workforce of a decade, or even, let's say, two decades ahead?

Maybe it's not even that far, maybe it's next week.

But what's a leader looking like being prepared for?

What kind of languages do they need to understand and be fluent in outside of, let's say, just English?

The good news is you won't have to learn any new languages and you'll actually be able to use languages you'll never have to learn.

So the, the T in GPT that it's called a transformer was originally invented for translation purposes.

And the example I like to use is one time I was working on an experiment with some German scientists and I was trying to be cool, this is 20 years ago, and I tried to translate it into German for them, but I used, instead of the word for computer server, I use the word for like a waiter.

And context matters, right?

Like, they did not.

That did not boost my credibility with that team.

So, like, they really cracked that problem with the context.

And, you know, you'll be able to talk, whether it's communication styles or languages, you'll be able to communicate with people and that you wouldn't have been able to do before.

Like, translation is just going to be absolutely insane.

Including context.

Like, the engineering team will say, oh, like what artifacts do we need for that?

This is something that actually happened to me today.

And I used Zoom's AI companion to say, like, what do they mean by artifact?

And it gave me an explanation in context that they were talking about something that people use on the website.

So you're not going to have to learn any new languages.

You are going to have to learn new concepts in terms of, in particular, in terms of shifting your ideas about what's possible.

So you can go, you can go launch a startup now with, you know, three people, can go launch a newborn startup because you can have an AI marketing person, an AI finance person.

You can also just a lot of people are using it in the obvious way, which is like, how do I spam more on LinkedIn?

Or how do I replace live chat reps, for example?

I think that's very short sighted.

I think that instead of, you know, one live chat rep, instead of doing three chats at a time, because the human brain can only do so much, maybe now they can do 30 because it helps them with the context and everything else like that, you could say, cool, I only need 1 10, the number of live chat reps, which is what most companies are saying right now.

Or you can say, great, my live chat reps can now spend more time digging into the problems and challenges of each person that they're talking to and create a competitive advantage for myself.

So you can definitely use it to lower costs.

But I think what's going to be most interesting is, you know, use construction.

For example, can I have it?

Ask the person a bunch of clarifying questions in advance so that I don't have to.

Can I have it, you know, translate the jargon that I'm using into jargon that the other people understand.

So you're not going to have to learn any new languages.

The mindset is going to change and then in particular the way that you think about your career.

So we were talking before this began.

Like I kind of started my professional career early 2000s, you know, the Internet was kind of like becoming a thing.

And because of that I got a huge jumpstart in my career, right?

I got to run HubSpot labs.

I had a CMO job when I was, you know, barely, you know, done not graduating from college, actually got to teach at the college I dropped out of because they didn't have a program covering the subject matter.

And for the, for the younger leaders, this is an incredible.

Again, my 16 year old nephew, this is an incredible opportunity where the only thing separating you from the people who currently hold the roles that you want to hold is that you don't have all the baggage you don't have, you don't know the right way.

And so because of that you're, it's much easier for you to figure out the best way.

All the rest of us, frankly, like I have to, I can't even switch to Superhuman for email, right?

I'm still using Gmail even though I'm trying to like use some of these AI tools just because, you know, three decades of habits, professional habits have trained me.

So it's a huge opportunity for anybody who's willing to incur the work to learn new skills and new mindsets.

Yeah.

Extraordinary.

You used some interesting examples before we jumped on that.

You've got AI summarizing your diary, using it, doing detailed research to, to kind of tell you about who you're meeting with, what they've said recently, what you might want to, you know, interesting things to ask them.

Like, you know, it's kind of worrying when you, when you say that to someone.

Oh, wow, okay.

That's like way more research than I was.

You know, I looked at your LinkedIn before we jumped on.

It's like, oh, wow.

Like it's, it seems so pathetic in many ways.

Right.

And that your, your approach is.

Did you say it summarizes it into audio and tells you that some of the key things.

Just tell us about, like some of your.

Because you're the kind of the cutting edge in many anyways.

Right?

Yeah.

And by the way, to clarify for everybody listening, I am not technical.

I cannot write code that is usable by anybody.

I can think logically and methodically and I can herd a bunch of tech cats, which is my actual job, is keeping everybody focused.

But one of the things that, you know, I hate showing up to meetings unprepared.

And so I have it look at who I'm meeting that day.

My Google calendar, it uses, I think it's Clearbit to give me all the information about your company.

Then it goes to LinkedIn and looks up you and your company's last, I think, hundred LinkedIn posts.

And then I have it synthesize based on what it knows about me, what are interesting things that we might be able to talk about.

So, for example, with you teed in on the young leaders thing because it's something I'm very passionate about and also just part of my origin story.

And then I have a generate audio because I go for a walk every morning and I could, yeah, sure, I could listen to the Economist podcast.

I could listen to, you know, the various podcasts that I used to listen to, or I can have it just also listen to those podcasts for me and only synthesize the things that it thinks I'll find interesting and just be really efficient in my day.

So I get a.

I get a President's daily briefing on everybody I'm meeting.

And if there's any interesting news about your industry or your company without me having to go do that.

That light amount of research before every meeting is a lot of time on our part.

And it's still, like I said, it's very, very shallow.

So now I can have these deep, you Know, briefings in context at scale, which is nice.

Phenomenal.

And I think the context piece is the most interesting part for me.

Sam.

Right.

Because you can put.

There's a podcast which you might want to listen to, and then you have to filter through and then you go, okay.

Like a friend of mine at McKinsey, funnily enough, sent me a podcast yesterday about developing leaders and their take on it.

And I was like, you know, I think it's a 20 minute episode.

I was like, oh, I kind of got a little bit stuck after the first few minutes because it wasn't directly relevant to me immediately.

But it's not to say the rest of it wasn't and there wouldn't be some gold in there.

But I think if I could have something listen to it and then tell me just like an assistant in many ways.

Right.

I guess is ultimately what we're talking about.

Listen to it contextually.

I well, given what you're doing right now, here's how it might be relevant and here's what they talked about and here's what you hear.

Some, you know, five quotes or something like that you could pump out.

To have that capability at your fingertips for free is just absolutely bonkers that.

It'S a literal agent, a literal assistant.

Everybody has their own ea.

Everybody.

I mean, I had IT draft a contract for me for the bank earlier this week and I still had the actual lawyer legal nerds bless it because that's more of a liability issue.

Like if it, if there's something in there is wrong, I want to be able to sue somebody who's not a non human AI, but they just set it back.

They're like, yeah, this is fine.

Right?

And so it's, it's, it's, it's giving you those capabilities.

It.

And like you said, most business books are the same too, right?

Like how much of the content we spend all of our time consuming.

Maybe 20% of the average business book is actually interesting and or new information.

For me, one of the things that I'm really interested in is this concept called information gain, where I want it to only give me things I don't already know.

I hate that, especially with the news.

Right.

It's just the same news over and over again.

But I have to listen to all these different podcasts just in case there's something I don't know.

We're quickly getting to where it'll listen to everything for you and turn 400 page business book into the 30 pages you'll find interesting.

Turn a 90 minute podcast into the 15 minutes you'll find interesting.

And then you can either like, I don't know, go play chess with the rest, you know, your free time, or you can just do more.

Like, I now listen to more podcasts than I ever listened to before because I don't actually have to listen to the entire podcast.

I can just get the bits that are interesting to me.

Wow, that is extraordinary.

I think one thing we talk about is the power of having mentors.

And, you know, and.

And often the junior people get stuck in the.

With a middle mentor that being like a very mediocre typically mentor at this level where, you know, the people they're surrounded by are people in the middle of their career who haven't necessarily reached the higher realms.

Not to say that the higher realms are superior in any way, but they have a very different perspective often about the right advice that you might get from a middle manager versus from an executive.

And yeah, I'm just thinking, wow, this is a real.

You don't even need the right network anymore.

In some ways, you could get.

The network is an AI.

You could have an advisory board every morning that you could interact with about, okay, here's the challenges I'm dealing with.

What do you think we should do?

What would be your different perspectives from around the table?

Is it that advanced yet in terms of could you have a CEO almost have like a conversation or an advisory board meeting?

It is very much, especially for the more senior folks.

So, like Dharmesh, who founded HubSpot, also founded Agent AI.

He fed it all of his keynote speeches, all the articles he's written, everything else like that, and created kind of a digital clone of himself.

So I actually do this on occasion.

I'm like, do I need to ask his permission for something?

And it'll be like, it'll give me some advice on that, right?

I'll be like, yes, you should ask for permission.

Or like, no.

Like he's.

He wants you to move fast and not bother with, you know, permission on things.

So being able to have kind of this counsel where you can get advice.

And there are some companies that, that just do this, by the way, that are much, much better at it than just a generic AI agent where you can feed it all of this stuff that I've written, all the lectures that I've given, etc.

And then if you're a startup and you're telling me you've got a retention problem, I'm pretty sure you actually have a pricing problem, not a retention problem.

If you're like a software startup and.

But I have to give that advice over and over and over again and because of that I only mentor a couple dozen startups and entrepreneurs.

If you clone me, I can mentor an infinite number of startups and entrepreneurs and also elevate those middle managers faster.

Right.

Tell it to give you a different perspective.

I always tell it to argue against me writing the contract.

For example, I'm like now pretend you're the bank and you're worried about me doing, doing something bad or weird like how would you change the agreement?

And it came up with a bunch of stuff that I didn't, hadn't even thought would be a concern for them.

So having it be multi perspective.

Right.

We always joke about explain it like I'm five.

There's also explain it like I'm the VP of sales.

Explain it like I'm the cfo.

Like the same piece of content should customize itself to the, to the person consuming it, but then also the same piece of advice.

Right.

I, I want to get advice from somebody who's Gen Z and actually understands how, I don't know, TikTok works or whatever.

Right.

Like I used to be cool and then time happened.

But you know, like can I, can I get like advice from very senior people, very educated and can I ask it to have a perspective where it's like, you know, what would somebody from Gen Z say about this?

Or are there other networks that I should look into?

Especially with some of the new stuff like the deep research models.

It's just, it's just fascinating.

I have it writing research papers for me on everything you can imagine and I learned so much that I hadn't thought to ask.

And those are always the most interesting questions and the hardest for mentors and why more experienced mentors are generally, I don't want to say better, but they just have more experience to do pattern recognition with.

Like they've seen this before.

Being able to have it go know what questions you haven't thought to ask.

I think that's, that's really interesting.

That's gonna be a big advantage for people who need mentors and not everybody can be mentored by the CEO.

Yeah, yeah, it's a great point.

I love that.

Really, really powerful.

I think if you can tap into it, maybe.

Let's finish with the last bit of advice.

Sam.

Let's say you've got a, let's say, let's say an engineer 2027.

They're early in their career interested about developing the right skills to put them ahead in the future.

What would be some of your guidance to them, let's say in construction infrastructure, if they're interested in being a senior leader down the track and being able to take advantage of the opportunities that we've got here at our fingertips.

Yeah.

So we're still obviously in the very early days of AI.

It feels like it's moving very fast and it is moving much faster than any other technology change.

The, the most important thing.

And I try to force myself to do this as well.

Just make it a part of your day.

Like, I don't use love.

My friends at Google don't use Google anymore.

I use Perplexity because I can ask it questions, it'll cite its answers, et cetera.

I the first thing you should do in addition to just like doing something dumb like I built a chess coach agent right where it just looks at my games and looks at my, the YouTube playlist of my actual chess coach and says like, what's he going to yell at me about?

That's fine.

But also just ask it to explain to you the lives of the people around you.

So, and you don't have to build your own AI agent for this.

You can use ChatGPT, perplexity, anything you want.

Just say like, here are the people you work with and interact with on a day to day basis and like explain their background.

Explain like, what is it like to be a project manager?

What is it like to be a comptroller?

What is it like to be a risk and compliance analyst and start to understand that AI can give you different perspectives and then start to think about how it can give you different skills.

There's a great site I love called there's an AI for that.com and if you go there you can just click into, you know, whatever job or task or industry you have and there's an AI for that, it turns out, starting to then say, what skills can I give myself that I don't have?

So I am not a good designer.

The last design tool I used was Microsoft Paint.

And so I, which doesn't even exist anymore.

So I use it to like help me design.

I am not.

I can write a thousand words, I can't write a hundred.

So I use it to edit things so that my essays, you know, to the team are actually just, you know, 50 word emails.

Figure out what it is that you wish you were good at and then go find an AI that can help you with that, whether it's data analysis or anything else like that, and then use it to Learn like, I have it just like the weirdest stuff, I have it going down.

Like I think I said earlier, you know, learning about quantum mechanics, learning about, you know, the.

The nuances of finance, like all these things that they're just helpful for you to have, like a broader context, but you would normally not have time to learn all of that.

And you need like a teacher.

And teachers aren't going to be replaced.

To be clear, TAs might be replaced, teacher's assistants might be replaced.

Because do we actually want human humans sitting there grading essays?

But I think that's my number one piece of advice is one, use it to figure out how everything you hate in your day automate it.

Like it's probably automatable unless it's physically picking stuff up where you actually do need robots.

Two, have it help you understand everybody around you.

And then three, have it help you learn and master skills that master skills without actually having to learn, you know, indesign in canva and design tools or data analytics.

Right.

Just be able to ask a question.

The people who win the future are going to be the people who are really, really good at asking AI the right questions.

Wow.

Super powerful.

So I'm scribbling like mad down here.

So that was asking for a friend, but mainly for me.

So, Sam, awesome work.

Thank you so much for your time.

Sam, I just want to maybe just.

What's one question around security?

Because I think it's.

We're giving a lot of information to the machines.

We're giving a lot of information to AI.

You're basically, you know, quote unquote, replicating yourself.

What are your thoughts around that?

Like, you know, the kind of the AI, Sam, being better than you, knowing more even.

I don't know, just especially in a free context, giving that amount of detailed information, quote, unquote, away.

How do you think about security and safety around that?

I have like, a specific punch list of ethical concerns that we're working on researching our way through.

One of them is just the consent of data, right?

So the big models, they vacuumed up the entire Internet, which is first of all unnecessary.

Like, you don't need the published works of Tolkien and the Lord of the Rings to write good LinkedIn posts.

It was unnecessary.

But then two, I also just think, like, you know, if you're creating content and putting it out there on the Internet, you should get to decide how it's used.

There are different language models that are being trained on things that, for example, only people have specifically opted in, or the data, the people who created the data have been.

Have been compensated.

So I think that's one bit is we're going to see a significant backlash.

And it's already, at least in the US it's working its way through our court system.

You know, if your AI is using, you know, somebody's book or movie or something like that, do you owe them something?

I would argue yes, but not a lawyer.

In terms of the data security, there's a huge move in all of computing to do what's called moving stuff to the edge.

So your phone has enough processing power to do most of the stuff that AI does without the data ever leaving your phone.

I think probably overkill, right?

Like, data security is actually pretty good.

We're pretty good at making sure credit card numbers and stuff like that.

Like, yeah, there's leaks.

But I think that in order for AI to gain adoption, the number one thing has to be trust, right?

So if I'm going to ask it for advice, if I'm going to be vulnerable to the AI, I kind of want to know that it's not going to, you know, that my boss isn't going to be able to log in and see that.

You know, I'm just not feeling it today.

And I need some motivation, which I actually use AI for a lot, motivating me to get out of bed and go run.

And I think while we can solve that as a technology problem, I just think that because it's such a human interaction and experience, like, if you were in the room with me, I would say anything to you, but we're talking through a computer, you know, kind of anybody.

Well, we're recording, so by definition, everybody's listening.

But even if we weren't, like kind of anybody could be listening in.

So I think that that is going to be big over the next few years as all of these AI capabilities.

And Apple is doing a very good job of this.

Actually not a huge Apple fan, but in this I am.

They're moving everything to the edge so that the AI models themselves never actually see the data.

Can you have the benefits of AI without any of the data that you rightfully own ever leaving a device that only you control?

I think that's going to be important and we're heading in that direction.

Interesting.

Very cool.

Sam, we've got.

We've come to time.

Thank you so much for your input here.

We're going to link to Agent AI.

It's.

If you're watching the video, it's on your T shirt, which is very helpful.

And you've said it a few times here.

So, yeah, huge thanks for your, for your help and your input on this one.

I think it's going to be a fascinating topic, and I hope people listening to this are logging in as we speak and checking out what you've got and how it can utilize them and benefit them.

So, yes, Sam, huge thanks for being here.

Thanks for having me.

And just everybody keep in mind that everybody else is just as confused as you are, but chaos is an opportunity.

That's a great way to finish.

I love it.