This week we are flipping the script a bit and our host is playing the guest to have Kyle James share his Builder Story. Kyle's journey from product management to podcasting is a testament to the power of curiosity and the belief that anyone can become a builder. Discover how his insights can empower you to embrace AI and transform your approach to building.

Kyle shares his unique perspective on the evolving landscape of AI, emphasizing the importance of starting small and iterating along the way. "Every person, every user can become a builder. They just need to understand that they can. This stuff's not hard," he asserts, encouraging listeners to demystify AI and take the leap into creation. Through engaging stories and practical advice, Kyle highlights the role of networking and community in navigating the AI era. His insights remind us that we're still in the early innings of AI, and there's a vast potential waiting to be explored. Whether you're a seasoned developer or just starting, this episode offers valuable lessons to help you learn, think differently, and take the next step in building with agents.

Learn more and connect with Kyle James:

• Connect with Kyle on LinkedIn: https://www.linkedin.com/in/jameskm03/ 
• Read Kyle’s Substack: https://kylejameslearn.substack.com/ 
• Kyle’s Guest Prep Agent: https://agent.ai/agent/prompted-guest-prep-research
• Kyle’s Post Episode Workflow Agent: https://agent.ai/agent/942s0tehcgr6ic07 

Subscribe for more Builder Stories and visit agent.ai to start building your own agents.

We've all heard the old analogy, kill two birds with one stone, right?

But the thing about it is, when you're doing something like a podcast, you can kill six birds with one stone.

These are all known pieces of content that we want to produce out of every single episode.

So it makes sense after organically figuring this out and doing this over some number of episodes.

Well, we can one shot turn this into an agency.

Think about how like a newspaper has journalists and they have editors, right?

And that's really what you're doing now as a human in the loop manager

of agents or LMs or whatnot is like, instead of writing the article, you're reading the article and then passing back the feedback of how it could be better.

I believe for an agent to work, it needs a lot of structure about what does great look like.

The only way you're gonna get to greatness is put the reps in.

Welcome to another episode of Prompted by Agent AI.

I'm not Kyle James.

Instead, today I'm interviewing Kyle James.

I'm Matthew Stein, the executive producer of Prompted, and today we are flipping the script.

We're going to ask Kyle about his origin story, how he came to start the podcast, and where his interest in AI came from.

What does he do on a day-to-day basis to explore the tools?

Why finding time to play is so important?

And what advice does he give to someone who's just getting started today?

We look at a couple of the agents he uses to run this very podcast right here.

So listen in as we pull back the curtain and let you see behind the scenes of our show.

Hey, Kyle, how's it going?

Doing all right.

How are you doing today, Matthew?

I'm doing great.

I like the fact that we're getting a chance to flip the script here because you're the host and we have listened to you talk on a bunch.

We're 30 episodes into this podcast.

Why should we keep recording more episodes?

Is AI dead?

Did I not get that memo?

Did somebody decide that AI is going away and not coming?

Is agentic not like the new buzzword that's even more, you know, bigger than generative now?

Like, we're the first inning, buddy.

Yeah, okay, all right.

So we've got a lot in the can.

There's a ton more to go.

You're right, there's a lot more runway here.

But in those first 30 episodes, you've talked to a whole variety of people, and we've really come a long way and told a lot of stories.

There are many more stories to tell for sure.

So how did you get here?

How did we start this podcast?

And tell us a little about that.

Yeah, this is kind of interesting, right?

Like we've all seen AI come big the last few years.

And I was in one of those positions.

Like you could see the job market changing.

You could see everybody freezing out, not knowing where things are going.

And me being the crazy wacko I am is like, you know what?

I'm trying to figure this stuff out too.

I'm learning.

I learn through talking to people and asking questions and figuring stuff out.

A lot of people do.

I was like, wait a minute, what if I could like figure out a way to learn, I love teaching and educating too, but also share all that content because at the end of the day, if AI does replace all our jobs, wouldn't the last person to be canned, the person who's helping everybody else understand this AI thing?

So that was like my thought process, right?

It's a little bit of like deceptive, like, hey, let me go do this thing.

And if it actually works out, then like, maybe I'll actually have some job security because people are still going to need somebody to explain it to them.

That's right.

We're always going to need the teachers.

So, okay, so you started this podcast, you were really curious about AI, how it's going to get used in the workplace.

Because that is really what we focus on for the most part.

Yes, we have fun.

but it's really about, how are people using AI in their jobs in the day-to-day?

And so you started with your first couple of interviews.

What are some of the things you've learned along the way?

Gosh, it's funny how people always say you just got to get starting.

And as someone who's been through a startup a lot, it's like 8 or 10 or something, you just have to get going, right?

Like,

You and I both knew like the first ones were not going to be the best ones.

And we had some people that were generous enough to kind of drag me along as I was figuring it out.

And no shades for our guests.

It's just tough to get started.

It's all me, right?

Like I'm very good at self-deprecating humor.

I enjoy that when I get uncomfortable.

So if I can make fun of myself, everybody else can too.

But yeah, there's been a bunch of themes that have really stood out to me.

And when we started, like I wasn't a builder.

I wasn't building Agentic.

I was posing because I was curious and wanted to ask these people to like help me learn.

And it was what of about 15 or so.

Errol, who's sales rep at HubSpot in Ireland, I think he's in Ireland, UK, somewhere in Europe, sales rep, like he didn't build anything.

But the way he was talking about how he was using the LLMs like ChatGPT, cursor, whatever, to help him write the prompts, like do the prompt engineering for him,

really clicked for me.

It's like, oh, wait a minute.

That makes so much sense.

I've been product.

I know how to build products, but I'm a failed software engineer.

Get a little bit of my background.

Like I'm a computer science person who hasn't major who hasn't coded in 20 years.

But like hearing him say that, like that makes sense.

Let me go try that.

And was able to write out the prompts that made sense to build agents with.

Like, okay, that's one interesting nugget of it.

another piece that really sunk in was, Redboard, both of you and I have a friend, Mike Redboard, really talking about the workshops and every single person who's ever attended any of those workshops came out as a builder.

That's true.

We run these workshops.

We actually still do.

There's a public ones that we've run a monthly.

But of all of the workshops, we've never had one where

people came out not having built an agent.

Yeah.

And I've watched some of them and it's like, okay, I got to do this.

I got to go build a couple.

And that's where we got some of our guests from too, because we've had some great builders on, some people who are just using the agents.

They didn't have to necessarily build them.

We've got tons of them that people can already use.

And I love hearing the stories of how these people apply these agents to their day-to-day life.

I think that's really important because that's

In the end, what everybody wants to know is like, how can I do this too?

And I think you do a really good job of telling that story, but aren't there other folks out here doing it?

Like what was your unique perspective that you wanted to bring about this?

What was the gap that you saw in the market to make space for the podcast?

There were people talking about AI, but I felt like I'm very much a storyteller.

I love, I learn through stories, right?

And I think so many people do.

To give you

A little bit of background, my dad was a Methodist minister, fifth generation Methodist minister.

And I remember the first time I spoke at a conference, this would have been 2008.

It was on e-mail marketing and my presentation was scheduled for 45 minutes and I went 20 minutes because I got through my slides, I put it out there and then I'm sitting up there on the stage like, oh my God, what do I do now?

I think we've all had those moments.

You didn't know what you didn't know.

Yeah.

And I remember going home and talking to him like, dad, you do this every week.

You get up in front of people every week and somehow have a unique, interesting thing to say.

He's like, son, it's like people relate to stories.

People want to hear stories.

And here I am telling a story about how to tell stories now.

But it really clicked for me.

And everybody's trying to figure out how to build AI and use AI and agentic now with agent AI was kind of a connection because

I worked at HubSpot 15 years ago with all you guys.

And it was like, here's an opportunity to tell these stories because it's, when you and I first talked about this, I fundamentally said, I believe every person, every user can become a builder.

They just need to understand that they can't.

This stuff's not hard.

They just need to find someone else who's done it for them that they can relate to.

So there's like a personal evolution that you're going through there, right?

Because we all started using the LLMs now to building agents.

Had you built any agents before you started the podcast or have you all done after them?

Walk us through that transformation.

I did.

I tried.

I failed.

I think the first one I really tried to build was just like, I'm going to write a cover letter generator.

Seems straightforward.

I think that's the big, you know this, the big trick about building any of these agents is what is something that you do many, many, many, many times?

And you've kind of got a clear understanding of what great looks like and what are the steps that need to go to do this.

So it's one of those things like take the job description, upload it, take your resume, upload it, give it a little bit of extra context about why you think you'd be a good fit, let it write a cover letter for you.

And it was just blowing up on me and it wasn't getting me to the right place.

And because I didn't know what I didn't know at that point about, gosh, maybe I need to have the LLM clean up my prompt a little bit more.

And so you're using LLMs to clean up your prompt.

You know what's good, but you're not quite there.

How do you get a little better?

It's an iteration process, right?

And I think, honestly, the thing for me at the time that I didn't really understand was working with the LLMs can get you there pretty much just as good as an agent, right?

But the fundamental thing that I think everybody needs to understand is, and I think we're going to get here in a second, is like how we're using, how I'm using agents to like produce these podcast episodes now.

You could use an LLM to do a cover letter, and if you just have a project somewhere and

the context loaded up, you probably do it 10 minutes.

But with an agent, you could take that 10 minutes down to like 2.

And that's the big difference.

Where before writing a cover letter would be like an hour.

So you could see where like, all right, 80% improvements.

Awesome.

Whoa, amazing.

Like, I don't need to get 95% improvement.

But once you've hit that 80%, you're like,

I want to go more.

I want to optimize more.

How do I get even more time and able to do this even faster?

And that's where building an agent out of it is that natural evolution, I think, for people.

So you're making cover letters.

You're on the job search.

I've also been on the job search.

We're in startups.

Startups are notoriously, I don't want to say unstable or fickle, but those are the two words that come to mind right now.

They're not always long-lasting positions.

So we've both been on the job search relatively recently in the AI era, shall we say.

How did you go from being on the job search to making your own job?

Because this isn't the only thing you do.

You have a bunch of irons in the fire.

I think like everybody else who was going through this, and this would have been 15, 16 months ago that I got laid off from a director of product role at a tech startup.

So for everyone who's listening later on, this is like mid-late 2024.

End of summer 2024, yeah, October.

And like everybody else, Logan and I explored a little bit of this.

It's like, first thing you do is you go start applying for as many jobs as you can.

This is some tips for anybody.

You're going to get so much more mileage out of networking than you will going applying for jobs on LinkedIn and everywhere else.

And all of the job interviews I had at the time and all of the interactions came out of that networking.

Nothing came from just blindly applying.

I mean,

Full stop.

I applied for over 100 jobs that way over a month, two, three month period.

One resulted in a screening call six months later.

Wow.

I mean, the AI, sorry, AI is just sort of like flipped job searching on its head.

So cover letters are great.

You know, they're very easy to create through agents or through LLMs, as we all know, but like

We've kind of killed it that way.

So networking, you're getting back to the, we always like to talk about AI plus human is where the real value is.

So you're doing some networking, figuring out whether or not you want to have another job.

Yeah, and I've got to eat.

I've got kids to feed.

I'm bored.

Like, how do I start making the side hustle?

And have done startups and started my own company before.

You just get to work.

And I've done a number of things.

I'm a research analyst now with a little firm, 360 Insight.

where we talk to go-to-market companies and do research.

I was doing some ad hoc project management.

I've been doing some product management for a local startup kind of as a fractional kind of role.

So I'm doing a lot of stuff right now.

And truth be told, there's no way I could be doing all that I'm doing if it wasn't for AI now because of the way it's able to supercharge you, right?

Like you could just work faster because you could get the first

We talk about this all the time, 80% of stuff done through an LLM and some basic prompting, and then you just have to clean it up afterwards.

You've done that with this podcast, because I've seen the agents that you've built to help handle the flow.

Obviously, when we're talking, when you're talking with the guests, it's very much human to human, people talking.

That's what people want to listen to, the folks that are listening to us right now.

That's the important part.

But there's a lot of

prep work in the beginning, and then there's a lot of post-work after the interview, not just the editing, but the promotion is a huge part of it.

And you built some really interesting frameworks around that.

Why don't you talk us through those?

Yeah, so I think the important thing for anybody to think through is they're turning something into an agent.

It needs to be something that you do all the time, and it needs to be something that you have done for some amount of time, right?

And why is that important?

Why does it have to be a task that you're doing all the time?

Because you really need something that, for an agent, I believe for an agent to work, it needs a lot of structure about what does great look like.

And if it's something that you're doing bespoke, on and off, like you don't necessarily know what great looks like until you do more of the crafting and molding, right?

But once you've created something a number of times, you have an example of greatness.

And I think that's, when we hear about AI slop, it's AI slop because it's not great, right?

Anybody could create something good, okay with AI.

Suno and Sora are loaded down with examples of good stuff.

But if you're going to put your name on it, it's got to be great.

So you have to know what great looks like.

And the only way you're going to get to greatness is put the reps in.

So you take these tasks that you're doing over and over again, because that's where you're going to find the efficiency too.

What's your process?

Do you start with a checklist of all of the steps that you got to go through?

The interesting superpower that I have is I'm really going to go from zero to 1.

Very much a generalist mindset because I've done so many darn things in my career.

So it's like, I don't know at first sometimes what it's going to look like.

You just got to got to start with something of like, I kind of want it to look like this at the end.

We knew we wanted to interview builders and tell their story.

But anybody that goes back and watch the early episodes, we weren't necessarily demoing agents then.

It's like that was an evolution that we figured out as we went along.

But I think the thing where so many people fail, starting a podcast or otherwise, is you just got to get started.

You just got to start doing something and learn and iterate and pivot and continue to improve

the process and you're upskill as you go.

And part of that is just like being your own worst critic and saying, every time you do it, like, how can I make this better?

Instead of just being content.

Yeah, and a lot of that, I think, comes from, I mean, you mentioned it, it's like, what does great look like?

And over the years, you've built and honed your sense of taste.

Tell us a little bit about where you honed your sense of taste, where you learned what does great look like and how to strive for that.

Podcast specifically, like I've been an avid podcast listener for over a decade, 15, 15 years now.

So I've listened to, I've consumed a lot of it.

was a, I would say it was a bucket list at some point.

Like, I want to be a podcaster.

That'd be fun.

How do I like nerd out and ask people interesting questions and like learn directly?

And it puts me in a position where I can have those conversations and share because I'm also, I love to teach.

I love to educate.

Little, little tick nugget for everybody out there.

I wrote part of the original HubSpot methodology on social media way, way back in the day.

That's not credited anywhere, but like, yeah, I wrote all of it.

So like, how do you combine all that stuff together and then listen in a certain podcast?

And I would say probably one of the biggest early influencers a few years ago into kind of like, oh, you could do business podcasting in kind of this tech product world was Lenny's podcast.

When he launched that one, it was like, oh my God, this is great.

And now, we've got Clairvaux out there doing like How I AI, which we are very much influenced by on this podcast.

We'll go for a little bit different audience and we're a little bit higher level where she's more hands-on, we're more storytelling.

But yeah, you see how people do things and you understand like, that was a great episode.

Why was that a great episode?

Analyze it.

And you could pick apart things like that and feed them into the LLMs to help you better understand some of that stuff.

So let's look back at the podcast workflow agents, because I really like the ones that you've built.

And I think people, our listeners, would benefit from understanding how you think about and refine the inputs and the outputs that you're expecting and going through some of that iterative process.

So why don't you take us through the first one there?

I think it's important to understand that when I first started podcasting, I didn't necessarily understand this either.

You know, I'd seen show notes and things like that, but I don't think I fully understand how important it was for the host to literally share with the guests like the list of questions and things like that you're going to ask.

And

my God, how do you come up with a whole list of questions?

Well, through iteration, I started building prompts that would work, that I could work with an LLM to say, here's the flow that I want to go through, the storytelling arc of like, what's your background?

Why did you do what you want to do?

Like, what became of it?

What has changed since you built this thing?

How can we help you?

A little more detailed than that, but there's this kind of general show flow that we follow on this show through all these episodes.

And kind of standardize that and

Build a prompt, build a LLM through a series of prompts that I'd written over time that lived in a notepad.

And I could take somebody and I'm just, I've actually loaded Brady here who, go watch Brady's episode, it's from a few weeks ago, but.

Brady's was a great episode.

Love that guy, great energy.

Use him as an example.

Someone who went through the workshop and like, I could build agents and started building them.

And so what we're seeing here is you using the agent that you've already built and you've run it, I don't know how many times, but

for each of our things.

And when we first started, this agent didn't exist.

So these were just questions, things you knew you wanted to pull together.

And now you've got it, it's down to just three inputs.

And so, theoretically, anybody could go and put together an agent like this.

And then, so your outputs, once they come up, talk us through the outputs and how you arrived at the format for those.

Yeah, so I would say a couple of things are going on here just to kind of tease the audience out is 1, this is chaining another agent that there is, you know, my agent team here, we had Tim Mackey a while ago on the pod, and he built this like podcast prep agent, which you could feed pretty much the same thing.

And this is why these are two of my inputs, because I need them for his agent.

And it goes and runs a background in person.

It gives you kind of their blurb, questions you could ask them, things not to ask, their areas of expertise.

It actually scores them on a scale of 1 to 10 about how good of a guest it thinks they will be looking at their LinkedIn profile and stuff.

So I take all of that and I run a few extra steps like, all right, here's his questions.

Now following this format, bring me back a whole bunch more questions that I can use.

What I've learned is it's better to come in with more questions than fewer, right?

And so I've got, what, six different sections here with three bullet, 4 bullet points each.

Guarantee I don't ask every single one of these questions to every single person.

But if it gets to a point in the conversation where I don't know where to go next, or I don't have a compelling way to do it, I've got something here.

And it shares with the audience kind of, or the guest, kind of what to expect.

Like where are the angles I'm going, what am I interested in?

It kind of sets the stage.

And it's unique to them in that conversation every time because

It's going through this series of things.

Is it perfect?

No.

Sometimes do I tweak some of the questions?

Yes.

But to this whole idea that the agent get me 80, 90% of the way there, one-shotted?

Yes.

And that's huge.

And I think that's part of the evolution of the 1st 10 episodes, I was doing all this stuff manually.

You know, the next 10 episodes, I had a series of LLM prompts that I was using and refining it as I went.

which would take the hour, let's say, of coming up with all the questions on my own down to 30 minutes.

Well, now that I can one-shot this with an agent, it's 5 minutes.

And that 5 minutes is basically me reading what this stuff is and tweaking one or two of them.

So this is great for a podcast agent, right?

But if we think about it, not just for a podcaster, right?

So like a lot of our listeners are not podcasters.

They don't have their own.

They have their own jobs.

They have their own things that they're doing all the time, but they can follow your format and then apply it to their day-to-day work life.

You're starting with pretty simple inputs.

It's doing a lot of research and summarization, and then you're getting structured outputs.

And that is something that you can take and apply to so many other tasks for people because it really just comes down to any kind of research project that you have.

Now, the flip side of that, there's

your prep.

I know you also have a post-podcast or post-episode agent that you run as well.

And that one you have, I see the results of that one.

So that one actually does get shared to a lot more people internally.

Not directly, you still do some copying and pasting, but take us through how you came to build that one.

Well, we've all heard the old analogy, kill two birds with one stone, right?

But the thing about it is, when you're doing something like a podcast, you could kill six birds with one stone, right?

You've got the podcast that you're producing.

You could produce, in this instance, we produce a post for the Agent AI blog.

We produce a post that goes to the community forums.

We produce

the YouTube video, we produce the audio podcast, we produce daily shorts the entire week that is there.

And we produce little short social segments that people on the team can use to post to LinkedIn and such like that.

So these are all known pieces of content that we want to produce out of every single episode.

So it makes sense after organically figuring this out and doing this over some number of episodes, well, we can one shot and turn this into an agent too.

Once again, it's really useful and important to remember, like,

It probably takes 4 hours to do all of this sort of content post-episode, right?

If you're working with an LLM and you know the general thing that you want to go after and what great looks like, you could probably do it in an hour to 90 minutes.

I know because I used to do that.

But now that I've kind of agenticized it, turned it into an agent,

You could take that down and do it all in 20 to 30 minutes because it's just a matter of reviewing all of the output and tweaking and cleaning it up some.

You really have to have that procedural mindset to understand your work, understand where the repeatability parts come in, understand where the human piece is still very, very important, and then figure out, you know, it's another tool.

It just helps you get your job done faster, more efficient, whatever it is.

I always think of before Excel, you would have a room full of people at desks doing math, and they were called computers.

And now each of those people was replaced by a cell in an Excel sheet.

And now we're all using spreadsheets all the time, and we say, God knows how many hundreds of thousands or millions of hours of manpower doing repeatable math.

by abstracting those into sheets.

So we've still got work that needs to be repeated, work that needs to be done sort of over and over again, and we're just finding more efficient ways to do it.

In some ways, it's completely mind-changing and game-changing and mind-breaking and whatever you want, but in other ways, it's really just the continued march of technology towards humans having better tools to be more efficient.

Yeah, and

Once again, if you have a workflow, a process that you do all the time, perfect to, turn into an agent.

At first, I didn't have that for this.

So it was more of like figuring out what that process was and how do you get more efficient and how do you iterate it to get it to a place where it is very repeatable.

So you've made a bunch of these, you've been talking to AI experts for, you know, 30 plus episodes.

Are you an AI expert yet?

You've been doing this for a while.

So are you an AI expert?

No, And once again, hey, that's why I wanted to do this.

But Jorge said this, Jorge Fuentes, a couple, that's like a month and a month and a half ago said this.

And I think it's so important to remember that the people that have been doing this the longest at this point have been doing it, what, three years, maybe four years since kind of ChatGPT first launched.

their first like LLM that people consume regularly.

Yes, machine learning and all that has existed longer.

And there are, you know, data scientists who know it more.

But for like use cases and all that, he summarizes so well because we're all toddlers at best, right?

Which I think is important to remember for two reasons.

One, no one has figured this stuff out that detailed yet.

And #2, you're not that far behind if you haven't.

Plenty of time to catch up.

It's kind of like this

I'm going to go way back in the story while this thing's loading, but Matthew, you remember Yoav Shapiro, right?

Oh yeah.

I remember having a conversation.

This was early first VP of software development at HubSpot back in 2008, 1910.

And I remember him sitting down and talking to me about like software engineering.

And we had been doing software engineering at the time for 20, 40 years.

relatively new in the grand scheme of things.

We have been building bridges for thousands of years, right?

So understanding what and how a bridge is built is pretty much solidified at this point.

And software engineering wasn't.

And here we are 15, 20 years later, we're seeing much more consistent frameworks and models and things like that to do it.

And now we're seeing the LLMs do it.

But like, it's important to remember that in these new

pastures for these new frontiers, like everything is being figured out now.

And especially with LLMs, everybody uses them fundamentally and completely different.

You and I were, I was just joking on a workshop that you and I were running last week that when I work with these LMs to write content, I intentionally tell it, remove EM dashes.

Everywhere that you would use an EM dash, go back and remove it.

But some people in that workshop loved you, like they write with EM dashes.

I don't.

So how do I make the content that it produces, that outputs sound more like me?

I intentionally go in and make it do steps like this.

So where do you find the time to play with these tools?

Because everyone's busy.

You got kids, I've got kids.

We've all got commitments and family and work things that we got to get done.

So how do you find the time to play with these tools?

Honestly, you do stuff like this.

And I don't mean like you do stuff like this to like find the time.

I mean, once you've done something like this,

and I'm able to save myself two to three hours per episode now, there's time to play.

You know what I mean?

So is it a chicken or an egg?

Maybe.

I'm going to go back to another episode because I learned so much talking to all these people.

We talked about what are you passionate about?

What do you love?

Go try to play around with AI and do those kind of things.

Whether that's you like video games or I like D&D or I like baseball.

You start building little models and agents to do some of those things.

And really, because the other advantage of doing it, things like that, is that you are a subject matter expert in those hobbies and things you're passionate about and love, right?

So you know what great looks like already.

You know how to identify it if it's hallucinating or if it's giving you BS and how to craft and clean it up.

Well, when you're crafting and cleaning it up, you're mastering it.

So then you've got a couple of use cases that had nothing to do with work or things, and you're able to say, oh, well, now I can take and apply this and do other things with it.

And now, like, I know how to do content output from an LM because I've got this use case that I did for a podcast.

And it really is finding those things, and you get more comfortable being able to do more and more, but you've got to just, yeah, find time to play.

And the hardest thing is how do you find time?

You take advantage of some of this to like give you a little bit more time.

I love the analogy of

people running alongside a bicycle that they don't have the impetus or they don't realize they should just stop and get on the bicycle and they'll start going much faster.

If you're running alongside the bicycle, you got to stop, get on, and then keep going.

And I think there's a lot of genuine fear out there all around AI, for sure, for lots of reasons.

But

One of the things that I see so much is like, oh, well, I have to do my work.

I can't stop to figure out a new way to do it by learning new tools or trying to build new workflows.

And I just have to keep doing what I'm doing the way I know how to do it because I can't make it go faster.

But one of the pieces of advice that I got when I was, many times that I've repeated to lots of other people is, when you're thinking about work, think about what's going to make your boss's boss happy.

And chances are, yes, you might be afraid of your manager, your direct manager saying, like, oh, why didn't you get to this?

Or why are you falling behind?

But chances are they are being asked to be more efficient.

They are being asked where can their teams save time by their managers.

This is an edict that keeps coming down from executives all over the place.

So if you can make your boss's boss happy by going to your manager and saying, it's like, hey,

I'm trying, I might be a little behind on these projects or reports or whatever this week, because I'm trying to figure out a way to make it more efficient.

They might be surprised and happy about it, because chances are they're being told they've got to find those things and you're coming to them with an answer they can use.

So, you know, if we all sort of like stop for a second to get on the bicycle, we can go a little bit faster.

And if you tell your manager, it's like, hey, I'm trying to do this.

Maybe they have budget, maybe they have advice, maybe they have internal resources who can actually help with these things.

So that's a really good way of, maybe you don't have to use your nights and weekends to figure this stuff out, but I know a lot of individual contributors and even managers who, if they had stopped and asked or just raised their hand and said, it's like, hey, I'm trying to figure out how to

how to use these tools a little bit better, they'll find they'll actually have a lot of support from management, from execs.

You remind me of an interesting story I learned years ago where one of my first bosses was telling me, it's this whole concept that we get promoted to the level that we're incompetent, right?

And it's one of those big dilemmas that people go through when they become, you know, a subject matter expert.

It's something that they're the best at, so they're the best at it, so they get promoted to the manager of it.

right?

perfect example of like you're the best sales rep, so you get promoted to sales rep manager.

So completely different skill sets, right?

Like as an individual contributor versus like motivating the team and whatnot.

And you've got to get out of your comfort zone a little bit to make that jump.

And this is a lot of that too, right?

Because you're going from being a subject matter expert, individual contributor, knows what great is, how to produce great and how to produce great consistently to like,

crap, I've got to let this thing now produce greatness and me be a poser, you know, claiming that I did it, even though it's kind of, you know what I mean?

It's really awkward.

And I think the best analogy that I've heard to really think about that is it's like, think about how like a newspaper has journalists and they have editors, right?

And that's really what you're doing now as a human in the loop, manager of agents or LLMs or whatnot is like,

Instead of writing the article, you're reading the article and then passing back the feedback of how it could be better.

And it's fundamentally different for everybody, how they think about it and how they react to it and getting okay with that.

How do you think about understanding market trends in AI adoption?

Because you do a lot of research and analysis.

So how do you look at the market trends that are out there and see how they start to get folded into people's day-to-day?

There's a couple of really interesting things that are happening right now.

One, we see

From the enterprise level, everybody's trying to lock down the security on it, right?

Cybersecurity is such a big, important thing, and also data privacy.

And how do you lock those things down into LLMs and what you can work with?

But here's the interesting thing, and you probably know that, I'm sure you know this, is like, the tighter you control the content, the context that's going in,

to produce the results, the actual better results that you get.

So the more you can lock that down and very tightly say, I'm giving you this little bit of data, make some kind of analysis, make some kind of predictions, research on it, gives you really good results because the context window is small and it doesn't run out of tokens.

So that's actually been really good.

The other really big challenge you hear from a lot of companies is like, everybody's going away from, this is going to take away jobs, because I think everybody's realizing like, we cannot pull humans out of the loop.

it's all about humans in the loop.

And you've heard me say it, and I talk to a lot of business leaders and go-to-market professionals that agree, it's like, get to 80%.

Like, can these things get you to 80%?

Because what happens at that point?

We're going to see more and more of professionals out there.

The relationship building is going to become the most important thing.

How do we build more and deeper relationships with people?

Because AI is able to take away so much of the grunt work.

I'll give you an example, right?

Like,

Think of sales professionals.

Like, have you ever met a sales rep who actually enjoys spending time in a CRM?

No.

No.

No, they don't exist.

They all hate it, and all the managers and all the RevOps people hate them because they don't feel like they put clean data in.

But guess who could be really good at doing that?

AI can listen to calls, it could pull transcripts, it could take notes, and it could get all the right data formatted in the right field so that all of the

data analysis people could do what they need.

And salespeople don't have to get yelled at for doing a bad job of that anymore.

While also getting the summaries and the agendas and tasks that they need to do later on and next.

So yes, they're absolutely, we're going to see a rebound into like relationship building and deep relationship with people.

It's going to matter.

And I don't think anybody's like upset about that.

Maybe some introverts, but you know,

Seems like a really good thing.

Yeah.

I think it's where, we talk a lot about, AI is not replacing humans.

It's that human-to-human interaction.

I could definitely see that in the future, being able to talk to a human or interact with a human perhaps becomes more of a premium or a luxury experience.

But especially when you're talking about, you know, business-to-business sales or something along those lines, people still feel reassured by talking to another person.

is that human-to-human relationship that matters, and it's everywhere.

It's in technology, it's in any other industry you mention.

We're not going to replace that with AI.

We will take away some of the drudgery of work.

I'm sure new drudgery will emerge, but that's just sort of the way this trend always is.

But I think, yeah, like in putting data into a CRM is a perfect example of time spent doing these things that could be more efficient.

So, if we're thinking about, how you get up to speed in your own journey, or how a listener can get up to speed in their own journey, what advice would you give to someone who's currently today where you were a year ago, like consuming AI content but not yet building with it?

What's the advice you would give that person?

It seems like it's such a cheat code, but you should be asking the LLMs, how can you do this better and faster?

And it'll kind of guide you through a lot of that.

Like I said, like I said earlier, like how I could use the LM to like write the prompts, prompt engineering for me.

I've done that trick a whole bunch.

Was recently putting together some case study content, pulling from call transcripts of interviews.

And it's like, hey, I've got this format I need for this landing page.

Can you take this thing and format it properly for this?

And went back and forth and cleaned it up.

And it said, how would you write a prompt to one shot this?

Perfect example of that.

Like, how could you work with these things to better think about that?

But I think as much as anything, it's just this mind shift that people have to go to through of like, you're not a poser.

You just are now the editor and no longer the journalist.

And that's a hard thing because, you know, you were the subject matter expert at this and now you're offshoring a lot of that.

But I think it's going to be a great thing for the age of generalists, though.

because people are going to have that ability to be cross-domain good enough in so many things because the LLMs can go deeper on things that they knew nothing about before.

And we try to sort of exemplify that by doing some building in public, learning in public.

What do you think about both the benefits and the challenges of documenting this learning journey?

Yeah, I think it's useful because

Other people come behind.

Other people need and want to understand the story.

The story is still what's so important.

You know this, and here's another plug.

It's like I have been writing on Substack for almost two years now.

One, because I could see the writing on the wall that I was probably going to be without a job very soon.

And like, uh-oh, I need to start getting some personal development out there.

But my journey started 20 years ago as a blogger in the higher education space.

And I was like, I need to get back into this.

For me personally, taking an idea and cleaning it up and putting it out there in kind of a published format is the equivalent of me turning in my homework assignment, right?

It's saying I've learned this thing and I could share it with you, tell me what I'm wrong about, beat me up, but this is what I've understood and put out there.

But also for me, it's also a way I can clean RAM out, like personal internal memory.

Like I've got this thing, I've mastered this thing, I could

dump it on the shelf.

If I ever need it again, I can search for it and find that article and refresh it or point somebody else to it.

I don't need to keep it inside anymore.

I could go clean out and start working on other stuff.

So I think for me, that's personally very useful and kind of how I think about it.

But also, I just love to teach because when I'm able to teach something, it's kind of that sign that I understand it.

Good writing is good thinking.

Yeah.

It forces you to clean up your ideas.

Exactly.

And I, still do, I publish on LinkedIn and I, anytime that I'm doing like thought leadership stuff, I'm almost never using AI for it.

Maybe a little cleanup, generally not.

Just because, yeah, it's like you said, like these are, you know, original thoughts or synthesis of other data, other thoughts that I've seen.

And the writing of it is almost my own exercise in order to get the stuff out.

And I can tell the difference in

the responses that I get and the feedback that I get when I am just using AI to spit something out.

And sometimes that's okay if I'm doing a standard publication or a standard publishing post for promoting an episode.

It's just important to get something out there, certainly if we're doing boilerplate things like reminder emails or whatever.

But when you're doing that sort of thought leadership stuff, when you're actually doing original thinking,

I think the writing is the important part.

It benefits the writer as much as the reader.

Oh, 100%.

And different people write and think and do things differently.

And I'll say another hack I've just started playing with in the last few weeks is actually using the record audio, like talking to the LLMs.

You read a prompt and I just will react to it in real time as I read it to like give it the next prompt.

And instead of typing it, because I don't, yes, I type very fast, but I could also react and say words even faster sometimes.

It may not be as clean, but it's something you could even play with is like, if you work in this model better, you don't have to write it.

You could kind of dump it out there and then the LLMs can like bulletize and make the hot mess that comes out of your head more crystal clear.

And then you can review it and like, is this what I wanted it to say or not?

So it's another way you could kind of work with these things and another method that people could follow that I'm surprised how useful and how good output I'm getting because I'm so used to like, I need to brain dump all this onto a piece of paper and make sure it's formatted and clean and clear to people where I could still do that, but I could speed it up a little bit and the LLMs can help me

crystallize and form what I'm trying to say faster.

Writers are not going extinct yet, for sure.

All right, as we wrap up, we've got our first 30 episodes in the bag.

It's been a great pleasure of mine to be the executive producer of the podcast and occasional guest.

And I'll be on a little bit more in the coming new year.

Kyle, any last advice you want to give for our listeners?

It's still a beginning.

We're still very much in the early innings of all this.

I think, you know, as you mentioned, we've got a lot of interesting stuff that we want to do next year.

diving into more of the educating piece, as long as telling more stories and diving deeper into individual agents too.

It's not too late to start.

I think that's the biggest thing that I've taken away from just talking to people and figure out how to have fun.

I was telling you the other week that I'm having a lot of fun right now with my kids, writing our own Christmas playlist with Suno.ai.

We're writing our own Christmas songs, which is great.

And it's fun.

And it's like, I said, find fun ways to play with it.

And you will learn.

You will learn how to write better prompts and how to create, I don't even want to say it's going to help me create better music, but it's like, it's just fun, right?

And sometimes the fun is where it's all at.

And I'm always known to say, if you're not having fun, what's the point?

You only get one life.

You might as well have fun in it.

Yes.

Right.

Kyle, thank you very much.

It was a pleasure talking to you.

I hope listeners enjoyed.

And we will see you on the next episode.