AI Knows Your Business Better Than You Do

you have this like intuition that you've built up, but when you hook an AI up that has the ability to sprawl across everything, it starts making not just two or three connections, it makes 4 5 6 7 8 9 10 connections. You can have another set of eyes, AI eyes looking at your workflow. >> Wow. >> I've got all of my work tools, Gmail, Calendar, Slack, Granola. Go look at all of this and pay attention to how I did my job. This thing actually knows my business. It actually knows it better than any individual in my company. One person can do all of those tasks. This changes the entire software industry.

[music] >> The Nick Stanley Show. >> Wade, welcome to the show. So excited to have you on here today. I thought an interesting starting point might be discussing your journey from Missouri to tech entrepreneur because it's an unusual path. >> Yeah. I didn't grow up as, you know, child entrepreneur. You know, you hear these stories of people who sold candy bars on like the school, you know, uh like at rec recess at school and stuff like that. That was never me. Uh, I grew up in a pretty like typical Midwest town. Like it was it was pretty conservative state capital and you know there was three jobs.

It was like you're going to work for the state, you're going to be a school teacher or you're going to be a doctor. Like that was what my exposure was. Uh, and so kind of just grew up like doing my thing like played music, was into sports, like you know that that was sort of the perspective I had. When things really started to shift for me though was in school. So, I was graduating around the time of the great financial crisis and uh you know, I'd been a good student uh and went to go apply for jobs, internships, things like that, but everyone had stopped hiring. Just what wouldn't wasn't, you know, even good students, they were like, we're not we're not interested in that.

And that was like the first time I thought, oh, I I'm not just going to be able to like cruise through this thing. I got to go like make my own way of this stuff. and uh was fortunate enough to find a small software startup in Columbia, Missouri where I was going to school. They were hiring somebody for marketing and uh I I had an engineering degree and you know I started to like study some online marketing stuff but I was not a marketer and uh the good news was they didn't know what a marketer was either. So, uh, for whatever reason, they thought I was a a good fit and they hired me. >> And that was like my first like real real upfront close seat to like what entrepreneurship actually looked like.

It was, you know, less than 10 people that worked at that company. Super entrepreneurial. We were all just trying to figure out how do we make a product that can sell? How do we get people interested in this thing? And and I fell in love. Like I I remember learning that one of our we we had people in Australia that were like finding the website, signing up for the product, paying us and and never ever talking to us. And I just my mind was blown. I was like, whoa, you can make money that way. That's insane that that works. And uh you know, I figured from that point forward, I was like, I got to figure out how to like start my own software company.

Now I came across Zapier in 2017 early on. I had a IT guy with a company that I started and he came to me and said, "Hey, you know all those things you've been asking me if there's a way to automate some of these repetitive tasks." He was like, "I figured it out." And and it was a real uh turning point for our firm in just becoming so much more efficient. And I wanted to just take a moment. What has Zapier been and what is it now as you seem to be cranking out new AI tools on a monthly basis? Let's just walk through that. >> Sure. We started as a simple way to integrate your tools. So, you know, if you're using things like gosh, in 2011, it would have been stuff like Dropbox and Evernote and Gmail and Trello and uh, you know, more B2B oriented tools like Mailchimp or Zenesk or Jira or whatnot.

And you could set up integrations between these and that's where it started. It was a very simple trigger actionbased system where an event happens over here, do this action over there. That's it. We quickly realized that folks wanted more than just one action. They want to do workflow. They want to do end to-end automation. And so that was act two of Zapier. Hey, let's actually build full-on workflow. And that came out in 2016 and demand skyrocket. That's probably where your IT guy got really excited about, oh man, I can automate this stuff end to end, >> right? where and just an example of that would be >> well a great example was hey I am running advertising on Facebook okay great when I get a lead I want to hit an enrichment API and you know get more information about the account and you know if it has this characteristic I want to save it to this list in Salesforce but if it has this other characteristic I want to put it in this other spot and you know if it's you know a hot lead I want to go text the sales rep and the sales rep should get notified right away to follow up instantly.

So stuff like that, you can sort of just think through in your head like what do what do I actually want to happen when I get a lead and how do I just make that just happen by default? >> Yeah. [snorts] >> So that was the next big shift of Zapier. I would say now we're kind of in like the third era or maybe like three, you know, 3B, which is the AI era. And it turns out both integration and workflow were phenomenal starting spots to build useful stuff in the age of AI. Integration was really it has been really valuable because for those of you who like use like agents and like push these tools to the max, you will understand this.

But for those of you who maybe just use vanilla chat GPT, you haven't yet experienced the magic of what happens when you hook AI up to your tools and to your data. If you just talk to vanilla chat GPT, you get kind of a generic response, a very intelligent response, but somewhat generic. But all of a sudden, if it understands, here's my business strategy, here's my, you know, customer ICP, here's what's going on inside my account. now answer questions based on that context. The experience goes from vanilla to holy cow. It's like this thing actually knows my business and it actually knows it better than any individual in my company because it's got access to everything.

So it's not just like talking to this one person who knows sales really good or this one person who knows engineering really good. It's like wow this thing has context on everything. So in some ways it's it's like a better version of you know if you assume the CEO is sort of has like the best you know perch of the business. The AI is even better because it can go all the way to the ground truth and read every last little bit. So the integration part super valuable to AI. And then the second piece the workflow side of the house well a lot of what these agents are are actually just workflows under the hood.

They're you know going through these steps. I got a lead from the website. I need to enrich it this way. I need to save it into Salesforce. That's what people are doing with agents anyway is they're using building workflows. Some of them are entirely deterministic. Some of them have like a little bit of like AI inference for certain tasks inside of them. Uh but it turns out all that infrastructure we built along plays so well in the age of AI. And so when you come to Zapier today, you mostly see Zaps, which is the old workflow product. You see agents, but you also see tools like Zapier MCP or Zapier SDK, which allows you to hook up your tools to your other agent platforms.

So, if you're using Claude Code or if you're using Cursor or if you're using codecs, you can bring all those integrations that Zapier provides directly into those tools and do them in like a very governed safe way. >> Yeah. And to just an example to illustrate all of what you're talking about, there was a video that we did on this channel that used a number of Zapier tools to make this happen where we did an integration with YouTube for this specific show where we and we built an AI board with some well-known characters Naval Ravocant and Seth Goden as advisors. And it was incredible what happened with the integration where the this chat GPT where the uh board was based could read all the transcripts of every episode and was directly plugged into the analytics of the show.

And when I started asking it questions about this show that I think I know everything about because I'm doing it every day and involved in it, it had insights into what was going on and what was working and what wasn't that I was not aware of at all. And it was a really incredible experience. And the video really resonated with the audience because and it just people popping off questions like how how did you do this? How did you make this happen? Um, >> yeah. And I mean that experience I think anyone who tries it the first time, that's their their their takeaway is they're like, "Whoa." You know, you have this like intuition that you've built up and your human mind has made connections to certain things.

But when you hook an AI up that has the ability to sprawl across everything, it starts making not just two or three connections, it makes four, five, six, seven, eight, nine, 10 connections across all this different stuff. and it comes back to you and you go, "Oh, wow. I never looked at it that way. I never thought of it that way. I didn't have the horsepower, the analyst capabilities, the the tools. I didn't have any of that to sort of surface that kind of knowledge." And it does change probably how you think about your show. You're probably like, "Mm, I ought to be doing more of this and less of that." >> Absolutely.

Absolutely. [snorts] Along those lines, what was the moment for you where it clicked that AI was going to be a real gamecher for society, for your company, for you as a CEO? >> Yeah, I think the the the moment where we took, you know, I I say raise raise our urgency level would have been the GBT4 launch. So we were dabbling with AI pre that. In fact, my two co-founders had started to build some AI product prototypes internally pre- chatbt. >> In fact, they built like a little texting bot where it was I they had built a phone number through Twilio and I could go text it and I would ask it questions and stuff and I remember asking uh what should I give my wife for her birthday?

I've got this weird top and like you know it's questions like that and it would respond back and give me suggestions. of course chatbt comes out and like that's basically the experience is this chatbot and I'm like ah yes of course like it makes sense that that should be a a product and yeah we were excited about that but we didn't necessarily pivot the business at that moment in time instead we shared chatpt inside of Slack we said hey check this out folks this is a cool product maybe has some implications on our space uh you know see if you can use it for new features or you use it to change how you operate.

But it was still very suggestive, organic, no no commands, no mandates, nothing like that. But then six months later, GPD4 comes out and there was three things that were impressive about GPT4. One is that it took six months to go from 3.5 to 4, so it was fast. Two, the model capabilities were just a lot better than 3.5. Just talking to it, it seems smarter. it seems it just it's you could see that like wow something something different happened here and then three the cost the cost had gone down quite a bit and so we looked at that and we said holy cow if we continue on a trajectory like this changes the entire software industry and so we called the code red internally around GPT4 and said hey we need to seriously rethink both our road map and our operations to take account what is coming with these these these new AI models >> and what is code red mean at Zapier [laughter] >> you know it's funny you ask that was what most we had never called the code red before so most of the team which is you know probably 700 600 700 people at the time were like great Wade code red what do you what does that mean because we never had one >> and so we had to figure it out on the fly you know I knew well it means we need to re rewrite everything.

We need to rethink all this stuff. It is like a sense of urgency, a sense of focus uh on this. And so we did a bunch of stuff, you know, we changed um a ton of things about our product roadmap. Uh we changed, you know, we bought a bunch of tools uh for people to start playing around with these things. But I'll tell you that there was one singular thing that probably had the biggest impact, which was we paused the company for a week and we ran a hackathon. everyone in the company, not just engineering, you got to go build some stuff with AI. And now, you know, this would have been still mostly just chatbt error.

There's not a lot of the tooling that you can find these days. And so for engineers, it was mostly working with the open AI APIs. For a lot of the rest of the staff, it was just like getting familiar with chat GPT and how it works and figuring all that sort of stuff out. But the results of that hackathon were we went from sub 10% of the employee base using AI daily to north of 50% in one week and this would have been in 2023. So that that was a huge step function forward uh at a time when most people were not using AI um all that heavily yet. >> Yeah. And I find the your interface that you've created for building agents is really ideal for the user who doesn't have a lot of coding experience or is not an engineer.

It is very easy to use, very intuitive. You're just talking to it and it makes things happen for you. Um, and I think that's kind of the the sweet spot uh where the Zapier agent creator is at right now. Uh, whereas maybe Claude Code can do incredible things, but if you don't have an understanding of software architecture, it is more horsepower than your average person needs. Even a a someone who has [snorts] some idea of how tools are built. Uh, Zapier's man, it's it's right there at the it's meeting the the need right where it's at. >> Yeah. I mean, that's what we've tried to do is make it really easy for folks who want to build these agents that run autonomously that run in the background.

Just just describe them in natural language. So, like the one I was describing before, it's like, hey, I got these leads coming in from Facebook. Can you help me enrich them? And then, you know, route leads like this here, route leads like that there. You can just describe like what you want and the system will like, great, got it. Let's go build it. um and it builds it out for you. Uh and it does it in a way that is reliable um costefficient um which is not always the case for these AI systems. These AI systems are not known for their reliability. They burn a lot of tokens. Uh and I think you know when you use something like Zapier where we're running it deterministically, we're encoding it in a very specific way.

You're going to get more reliability and cost advantages out of it. >> [snorts] >> Well, one thing I like about it that is underestimated I think right now is that it will also tell you when it would be useful rather than having an agent that is trying to make judgments and decisions to just go with a deterministic automation that just says, "Hey, when there's a new lead, I want it added to my HubSpot or whatever that might be." >> [snorts] >> And I found that almost ironic that some of I'm I'm getting on there to use the AI functionality and then it's telling me, hey, just use the old school Zap that will just make this happen automatically.

I'm going, why didn't I think of that before? >> Well, I'll tell you a secret. So, a lot of the value of working with AI is to help you build the tool. So, you know, when you're trying to figure out like what should my agent do? What should this workflow do, it is really helpful to talk to an AI to an agent system about what you want? Go back and forth, design it, and think through all that stuff. But once you get it working, then you're like, "Okay, great. I want you to do that exactly that same way 100% of the time. I want you to run a plain old deterministic workflow because that's what I want." And you know, I think where a lot of folks make mistakes is they actually build it with AI, but then they also run it with AI where the AI's making decisions every step of the way and it's not always running it the same way you want.

Uh, and so a lot of the times the value of AI is actually in the build process. On the running side, run it with code. That is a way better way to go about it. >> Yeah. now and I've seen with your hiring, you're really putting AI fluency at the forefront with your hiring and I want to understand why that's important to you. And then do you see this AI fluency as something that is lifting up the floor on people that you're hiring or is it really amplifying your top talent or is it doing both of those things at once? >> Yeah. So AI fluency is important for us for a couple reasons. One, when folks really understand how to use these tools, we reap the benefits internally from them.

So our operations get better, our products get better because we've got talent that actually understands how to get the most out of these tools. So that's really important. The other thing that's really important is it's a signal to the market that Zapier cares a lot about this stuff. We invest a lot of these things and I think right now that helps us attract the best talent to Zapier. Unfortunately, there's too many people that I come across that are excited about this stuff. They're building cool stuff in their spare time. Maybe they've got an open claw setup. Maybe they've got, you know, this that or other.

But they get to work, everything's locked down. Everything's shut down. they can't use AI for this, they can't use it for that, they can't they they they effectively can't do anything. And so these folks who have started to build these skills and capabilities, they can't really, you know, it's like you got a Ferrari in the garage. It's like, why why is it in the garage? Get it out there and race that thing. Um, and that's happening in a lot of work forces where they've got this amazing talent internally, but they're not actually letting it get out on the racetrack and and go make things happen.

And so those folks start to see what we're doing and say, "Well, if I if I got this zap here, I could I could be out on the racetrack. I could be be making stuff happen." So those are the the reasons that I think, you know, this AI fluency is so important for us. Then you ask like, hey, is it is it raising the floor? Is it raising the ceiling? And the answer is it's it's doing both. um you know you I I would say the the thing that happens first when everyone builds with AI is certainly everyone's individual productivity starts to go up. So you can see across the company you know just you can almost ask anybody you can say like hey you know do you feel more productive with AI and they'll all be like yes like I used to do it this way I do it that way etc.

I think where the ceiling raising starts to happen and this is where I think a lot of the organizations are that are have been fast to adopt AI. This is kind of the next bottleneck that we're all trying to fix is how do we go from individuals being great to like institutional productivity going up, right? That's the next big unlock. Just because I'm, you know, you know, if I'm better, but you are the bottleneck in the org, it doesn't matter that I'm now 10 times faster, 100 times faster. If you're still in the same spot or the system in which we operate is still at the same spot, my acceleration doesn't help the the company.

And so, you know, I think a lot of folks are trying to figure out, okay, how do we actually reinvent these endto-end systems? How do we reinvent the org chart? How do we reinfor gets done to actually take advantage of that? And this is where, you know, it's it's people, it's tech, but it's also redoing a lot of the sort of fundamentals. And so, um, you know, I think that that's kind of the next frontier it feels like for a lot of these scaled organizations. Now, you have a unique viewpoint with your firm to I feel like to see a little bit into a crystal ball that not all of us have because you can see how people are using these things and and what they're experimenting with and what they're making with the tools.

Where do you think all of this is headed with AI in general? Well, I I think the exciting thing is that an individual can now enact so much more like can solve so much more of the problem. You know, it used to be, you know, the way a company would work or like at least a software company like which is what we we do is, you know, you'd have a team that goes out and talks to customers, understands the market problem. They would go, you know, share some of that stuff with the product team. The product team would shape the road map, figure out what to do. That would get handed off to engineers and designers who would go build all these systems.

And at the end, you would have like a test release cycle. And then you would have marketing that would do all these things. And so, you know, there's all these people in this chain of work that sort of has to go through to to get something out the door. Now, with AI, one person can do all of those tasks, every single one of them. And I think that that is super exciting. Like, it's going to be so interesting to see what the impact of that is for society. I think one, we're going to just see far more um a longer tale of products that get built for consumers because the cost of producing it has gone way down.

And so that means we're going to solve more problems that didn't used to be economically useful to solve. I think you're going to start to see that happen inside of companies. We're already seeing this now, too, where I've got, you know, I don't know how many agents running for myself that are solving problems that I always wanted to solve, but I never could. I just never could solve them, but now I I've got them running and they're super helpful. Um, and so that's like very exciting to see. Um, ultimately though, I I do think that we're going to see like a big rewrite in how just like companies operate.

Probably the probably if I was to point to one particular thing though it would be the shift from co code writing code being expensive to writing code being cheap. So much of our workforce or our systems and processes were baked around this principle that code is expensive to create. So you do all this upfront work, you do all these things because once you start writing code, you know that you better get it right. It's gonna be expensive to write all this code. But now if code is cheap, how is the world different? You don't have to have all this upfront investment. You can do a lot more just like, well, let's just try it.

Let's just build it and see. Let's see what So all this upfront coordination, effort, all that sort of stuff. I think a lot of that fades away and now you've got this other thing that's emerging where it's like, well, we're building a lot of stuff. Well, do we ship it all? Should we ship it all? Uh, you hear taste come out a lot. You know, some people are like, "Oh, taste is the new thing." Maybe maybe you're just shipping a lot of things and you're AB testing every stuff and like the things that win are like staying and the things that are losing are getting rolled back and taken out of the product.

Maybe that's what it looks like. Um, I don't know. It's it's just really interesting to think about like how the world and the organizations, particularly software organizations, shift when code is cheap versus expensive. This episode of the Nick Stanley Show is brought to you by Zapier. If you've ever felt buried in repetitive work, copying data, moving files, sending follow-ups, you know it's like death by a thousand mouse [music] clicks. Zapier has always been the tool that fixes that. It connects over 8,000 apps, Google Drive, Slack, Notion, Gmail, MySpace, you name it. So your tools can finally play nice [music] together.

But here's the big shift. Zapier now lets you create AI agents with their chat GPT integration. Think of them as tireless teammates who never complain, [snorts] never take lunch, and never get bored of doing the boring stuff. I've actually made several of these little agents myself. No Python, no JavaScript, just pure vibe coding, which is my favorite kind of coding because even though I have no idea how to code, I just talk to the AI, tell it what I want, and almost magically it works. For example, when a podcast guest books a time, Zapier can send them a personalized email from me preparing them for the show, create a draft of show notes, update my calendar, and even prep a social media post automatically.

All of which frees me up to focus on the important stuff, like taking credit for the amazing work my agents did. And now, Zapier just took it up another level for developers. Zapier MCP is available right inside Chat GPT, which means you can connect those 8,000 plus apps and trigger workflows just by [music] writing what you want to happen. Literally, you tell ChatGpt, "Send this file to my team in Slack, update the spreadsheet, and draft an email, and Zapier plus ChatGpt figure out the right tools do it for you." Getting started is simple. Head to the Zapier ChatgPT MCP server and add the tools you want ChatGpt to access.

Follow the steps in the connect tab. And if you're an admin on a Chat GPT enterprise account, you can enable MCP across your entire workspace. So, if you're ready to stop wasting time on busy work, [music] join the AI revolution and make a little automation magic of your own. Try the Zapier chat GPT integration using the link below. Yeah, you talk a lot about contact with reality as being a great test for a new idea, a new product, whatever you're working on. And I think if you are an entrepreneurial kid in Missouri, this is a really or wherever you might be an exciting time to be alive because you could just make stuff and put it out in the world and see what happens.

>> Yeah, spot on. Like I I think that this is got to be one of the most exciting times to be a startup founder because you know there there is so few barriers to getting started. So few you can literally just make something and you know you can put it online. You just say like hey you know check this thing out and if you've made something truly useful people are going to check it out and they're going to try it and they're going to tell you and they're they're going to like it if you've done something truly useful. Uh, and so and if and if you haven't, you'll you'll learn that, too. And so you can like start the feedback loop and go, okay, well, what's the next thing I got to go build?

Let let me try another thing. And the the feedback loops are just so much faster than they used to be. Uh, and so it makes it really exciting to build. >> Yeah. So if you were, let's take three age points. If you're 13 years old, if you're 23 years old, or if you're 53 years old, what are the things you would be doing at those stages of life to be preparing for what's coming next to get involved in what's happening now, the skills you would want to be building, whether that's a a kid or a executive in a firm. Well, you know, I think all three of those age groups benefit from just using AI, like learning how to wield these tools to do interesting stuff.

So certainly I think everyone benefits from building some amount of AI fluency. Now maybe what's different, you know, if you're 13, you know, you still got a lot of education in front of you. And so I still think there's probably a lot of value in building out just like a very well-rounded education. Like you you ought to be learning like you know math and physics and you know communications and like just all these sort of like core foundational skills like those things are going to be critical for you no matter what like you end up doing in your life. And so, you know, I think in addition to just like, yeah, get good at using AI, get you good at using these tools, like there's so much you just want to build your like foundation of knowledge and these these skills that help you understand the world um really well.

Now, if you're 23, you probably developed a lot of those things. You're entering the workforce. You're coming into the workforce in a really interesting time. You know, I think probably the bad news for you is that you may not have been prepared for this AI world. Um, you know, you probably went through college and got a degree in a world that didn't know that this technology was going to exist. Um, or it it has existed, but it's been so new that nobody knew what to do with it yet. And so, you probably have an education that maybe is only partially relevant for the world that you're coming into.

Um, so I think that's the bad news. Well, and also the bad news is that a lot of companies are like putting hiring pauses in place and like, you know, they're trying to say, "Hey, we got to take stock before we go hire people. We need to understand what it is we actually need. So, let's just take a beat real quick." And so all these orgs that were so used to hiring 22, 23 year olds all over the place, they're maybe saying, "Ah, let's let's wait." So that's the bad news. I think the good news though is that you have you're at the very beginning of your career. And so like you you have a blank canvas.

You have all this time on your hand where you can actually go learn how to be the best at wielding these AI tools. And what is unique about this is that the folks in the workforce, the folks that are 30, 40, 50 years old, they are also at the starting block of using these tools too. And so usually when a or you know you know 23y old comes into the workforce like someone who's 33 or 43 has that many more years of experience form they have a head start on this stuff. Now not so much. Like they're kind of at the same starting block. And so the good news is there's never been a better time for you to say, "Hey, I'm gonna like I'm gonna race ahead and get really good at these things." And I might actually jumpst start my career compared to everybody else in the workforce just by getting really good at using these things.

So I think if you're 23, you probably are in the best position of your life to actually go learn how to wield these things um and do impressive stuff with it. Now, if you're 53 or whatever the age 50, I think is what you said. What should you do? Well, I've got about another decade before I get to that age. So, you know, I I I might my answer probably not less good for this because I haven't experienced that quite yet. But, you know, to the extent that you want to continue working, I think you got to go learn learn how to use these tools. Um, you know, my, uh, kind of reminds me of my mom's a pharmacist and she came into the workforce before computers.

Uh, then eventually computers showed up. It's like, well, if you're going to be a, you know, if you're going to work in medicine, you're going to have to learn how to use all these computers. You aren't going to get to do it the pen and paper way anymore. Well, I think this AI era is going to be the same thing. Like, if you want to stay in the workforce, you're going to have to like keep your skills relevant. And so, time to reskill all this stuff. But you probably also have a really great opportunity to think about, you know, what your legacy looks like, what your stewardish looks like. Like you've got so many so many great experiences from just being in the workforce for so long that you probably can go look back on all of the things you've experienced and think what how can I actually go solve those things with AI?

how could I have done do something better? Uh, and so you're probably in just a really good spot to have more ideas because you've just seen more stuff than most people. And so if you actually like tap into that part of your brain, I I think you can probably actually do some pretty pretty incredible stuff. Um, yeah, I think the folks further on their career, the the worst thing you could do is just sort of, you know, stick stick your, you know, head in the sand and say, "Nope, I don't want to I don't want to see ch see change." Um because I think change is coming whether you like it or not. >> Yes.

Yes. [snorts] Along those lines, do you still believe in the value of higher education? >> I I do think that higher education can be incredibly valuable. But that is a complicated question. You know, I think the parts of higher education that are really valuable are, you know, if you're learning, you know, if you're if you're learning things like physics or like uh math or these like hard skills, like those things are genuinely quite useful. I also think higher education if used appropriately, I don't know that I totally did this well, but if used appropriately, it creates a space for someone to actually go explore and figure out what their interests are and you have this sort of space to like dabble in a bunch of different things and, you know, kind of figure yourself out.

Um, you know, you can let your sort of curiosity take you um to what is most interesting. you know, higher education has failed us. Is that, you know, we've got kids that are taking out just huge huge loans and, you know, getting into industries that just there's just no way that they're going to be able to pay those things back. Like the ROI on that just doesn't make any sense. Uh, and so yeah, I think there there is definitely going to be a reckoning on like that side of higher education, but like higher education as a concept, I don't know. I still like a lot of it. I I I benefited a lot from my higher education, but there's other parts of it that I look back on and think kind of wasted my time, too.

So, it's it's very much a mixed bag right now. And I think there's increasingly growing camps that are saying, "Hey, like we're we're not getting the bang for the buck on this stuff." >> Yeah, that does seem to be a strong sentiment right now. And I feel like colleges are in a place much like companies where they if they get on board with this and start integrating teaching skills and these tools for the future, then they have a future. And if they put their head in the sand and just kind of continue on as they have for the last 50 years, they will get left behind. [snorts] Is it I can't remember if it's Arizona University or Arizona State University, but I saw that they they just announced starting next year for all freshmen.

They're going to have an AI research course that everybody has to take, no matter what you want to major in just to get everybody moving in that direction right from day one. And then a almost like a dual major path. So you could major in political science or physics or whatever it might be, but have this dual major in machine learning to implement that in whatever you get into. And I thought that's what the future of education for the colleges that survive might look like something like that. >> Yeah, you've got to give props to anyone that is willing to try something new and novel out. Like I certainly think the the worst strategy is to stand still and you know most higher education institutions are probably going to stand still.

>> Yeah. Yeah. Um now you something you you mentioned I'm going to paraphrase here but was that you should try to automate part of your job away. And what do you mean by that? Well, I basically means like you've you've got a bunch of tasks that you do every day, every week, every month, etc. You should find ways to automate as much of that as you possibly can and then redirect your time and attention to the next big problem, the next hard thing, the next thing that's going to generate massive amounts of ROI. Um, I'll tell you the one of the fun loops I've gotten into is I use cursor as my agent of uh like driver of choice and I've got the Zap year SDK hooked up to it and with the SDK I've got all of my work tools.

So, Gmail, uh, Calendar, Slack, uh, Granola, which is my meeting recording tool, uh, so on and so forth. And every week on Saturday, I have an agent run that runs and I say, "Hey, I want you to go look at all of my cursor sessions, all of my emails, all of my Slack threads, all my granola. Like, go look at all of this and pay attention to how I did my job. Where can I build tools and automations that will either make me more efficient, deliver higher quality, or reduce the time that I spend on these things. Uh, and give me suggestions. And every week it comes back to me with like half a dozen to a dozen ideas.

And usually I have them go build probably about half of them. And so every week I'm finding new ways to automate just just little bits and bits and bits and bits of my job. Probably my favorite one that I've done is at the end of every day, usually I I've got a pretty meeting heavy schedule. Uh talking to customers, internal meetings, reviewing work, etc. So at the end of the day, I usually have tons of just like follow-up stuff I've got to do. I got to send an email to intro this person. And I got to post a quick status update on this project back to Slack. I got to just like there's a long laundry list of these things.

And it used to be that it might take me an hour, maybe two hours to just like work through all the like status updates and follow-ups just from the day's work. Well, now I've got an agent that's listening to all those granola transcripts. And so at the end of the day, it goes, "Hey, Wade, your draft emails, those are already drafted. You just go take a look at them and you press send if you like them. By the way, all those Slack updates you need to send, I've already drafted those here. You take a look and you copy and paste them and send them over here. And so what used to be like an hour, maybe two hours, it's down to like less than 10 minutes.

>> Wow. >> And so that's that's the kind of benefit you can get when you're just thinking like where can I just automate more and more of my day? Uh, and the good news if you use that tip I described, which is like, hey, go look at how I'm working, you can have another set of eyes, AI eyes in this case, looking at your workflow. It's the same thing that you asked, you know, uh, before when you were saying, "Hey, go look at my YouTube transcripts and my analytics, and I just want you to go see what I'm not seeing." Um, and that that is such a powerful hack to help you come up with ways that you can be better at your job that you may not you may not know, you may not recognize um that there's an opportunity for you.

>> Yeah. And just to make sure everybody understands, so with Granola is recording all of your meetings and that's an AI tool that will summarize things, but then >> has all the transcripts. Yep. >> Right. And then you're feeding that into your agent along with emails and your calendar and maybe a couple of other sources. And then it is drafting some of those responses for you because it can see, oh, Wade, you're supposed to connect Nick with the next awesome guest that's going to be on his show or whatever it might be. And then it actually drafts that email for you and it's sitting in your inbox.

>> Totally. So, you might say at the end of this episode, you'd be like, "Wait, I loved having you. Um, who else should I have on the show?" And I'm like, "Oh, you know, my buddy so and so would be a great guest for you." And be like, "Well, would you would you mind doing an intro?" Great. So, you know, that's in the granola transcript. My, you know, it my agent picks up on the fact that like, hey, Wade needs to send an intro to that. Feeds it off through the Zap Your SDK to Gmail and it drafts the the email. So, you know, I'll go at the end of the day, pop open my drafts, and I'll say, you know, and I'll just check it.

They're sitting in my draft folder will be an email to Nick, uh, saying, "Hey, Nick, thanks for having me on the show, CC to soand so um, you know, you wanted to have an intro. Here it is." Uh, you know, hopefully you guys have a chance to connect. >> Yeah. Well, I am really enjoying this conversation, Wade. And if you have some [laughter] great ideas, uh, have have your AI let me know. Um, now at this point, where do you see it being critical that humans stay in the loop? Right? Because when we're talking about agents making choices for us, that can be a tricky part to figure out. >> Yeah. the pro probably the um the best articulation uh of this that I've heard is this concept of the lethal trifecta.

So the lethal trifecta is when you have an agent that has three things. One, it's got access to private data. So that could be like my email inbox, you know, my the company CRM, you know, things like that. Uh it's got exposure to untrusted content. So this is, you know, someone from the outside world can like, you know, prompt send a prompt into it, can query it, can do something interesting to it. And three, it's got the ability to externally communicate as well. So that gets really dangerous. And this is where when you hear people talking about hey there might be a cyber incident with agents etc.

they're usually concerned about when these three ingredients are there. So, for example, if I set up an agent that you know our uh well, actually a great example was um this has been I think maybe a year ago now was uh Air Canada where I think it was where somebody talking to the chatbot got it to like issue a massive refund or like buy buy a flight for like pennies on the dollar or something like that. And that's because like a third party could go interact with it. It had the ability to do something interesting and it had access to all these things and so it caused the agent to make a decision that a human would look at and go like no no you would never do that but it had all those capabilities and so you had kind of a a situation there that caught you know got got headlines.

So right now to avoid that basically what you got to do is build agents that only have two of the three. As long as you have two of the three you're you know you kind of have a safe agent. Um and there's a lot of like discussion in you know AI product development etc where it's like hey is it possible even to like have all those three ingredients together and have it be safe. >> Um that best as I know that is an unsolved problem >> right >> to to this point in time. So there's a lot of people thinking hard about it. Now, what I tend to recommend for folks is like, hey, even still, if you're building one that has two of these three, it is useful to have a human in the loop in at least initially.

So, for example, this example that I'm talking about where I have this agent go draft emails for me. Well, you can notice I'm having them draft emails, not send emails. And so, that's a mechanism of inserting myself as a human in the loop. Now, it's possible over time that, you know, I find myself 99 times out of 100 never making an edit to that. Well, if that actually emerges, maybe I say, you know what, actually, just go ahead and send the email. Like, if it makes one mistake in a 100, I'll live with that because it's right 99 times of of 100 and that just saves me a bunch of review time. And it turns out that example I just shared there actually mimics what um manufacturing facilities have done for a long time which is they have this process of statistical quality control.

And so any widget you see get made in uh society generally follows this this process where you know manufacturing mi machine will just has a has a quality control standard and they'll you know stamp out you know hundreds thousands of widgets at a time and the way in which they keep quality high is they usually sample completed widgets off the line and so you know if a hundred widgets come through they might sample one or two and based on statistics they know exactly how many they need to sample to say with statistical confidence, this batch of widgets is all good. And so I think we're going to see a lot of systems like that get built up where, you know, hey, we actually might have things that are very very automated or mostly automated, but humans will be sampling them and based on statistics, we'll be able to say, hey, enough of these were high quality that we're we're confident that saying within this like error band, the quality hit this certain threshold across everything here.

And so I think that's how we're going to see these systems go from mostly human in the loop to only partially human in the loop is we're going to use statistics to basically tell us according to a certain quality control threshold that we are comfortable automating big chunks of this task. >> Quick pause. This is important. There are only three things you can train in life. Your craft, your body, and your mind. Most people work on the first two and just hope the third one shows up when it matters. That's why I'm recommending Finding Your Best by Michael Jery. I've done the course myself. It's excellent.

It's not motivational, not fluffy. It's structured to train the skills that actually drive performance. Confidence, recovery, trust, grit, clarity under pressure. If you ever feel capable on paper but inconsistent in real life, this is about closing that gap. I don't put my name behind a whole lot of things. This one earns it. You can find the link below. No urgency, just a genuinely solid tool if you're serious about getting better. Now, something people are terrified of, whether it's within software or industries across the board, is they're wondering, is my industry one that's going to get quietly hollowed out and then I'm going to be a victim of that?

What industries or job types do you see being most at risk for that? Mhm. Um, good question. You know, I think the most obvious one, and these these companies have already felt it the hardest, is places that are just purely information driven sites. Um, you know, the one in the tech industry that I think a lot of people keep tabs on is Stack Overflow. So for the longest time there's a site stack overflow where engineers would go to ask questions and you know the community would respond and you know say hey why don't you solve it this way or that way >> right >> and it turns out that it's a lot easier to just ask coding agent to just go talk to to just figure this stuff out so people don't go to stack overflow anymore.

Uh and so that's a that's a product experience that's been hollowed out. There's going to be a lot of these types of things where it just turns out like the better way to solve the problem is talking to AI rather than you know this other product or service. And the second part of this question you said is well what about jobs? You know, I think that there will definitely be jobs that are changed quite a bit because of AI, but I think individual humans are much more malleable than companies. You know, companies are, you know, require steering hordes of people to like reinvent themselves. A singular human can just choose to upskill themselves.

And so you know in the case of a human you know you might see that or we we're actually actively seeing this where take engineering as a profession. Many engineers are not writing code anymore. The demand for engineering talent is growing. There's more open job listings than there ever has before. Well how can that be? like you know engineering has writing code has effectively been automated now but yet we're still hiring all these engineers and the reason why is it turns out that the job of engineering was never to write code that was a task that engineers did but the job of engineering was always to solve this problem to build this system to do this thing and so as a result because it's a lot easier to write code now we're finding well we can solve more problems and it turns out engineers are actually pretty good at using these systems to solve more and more and more and more problems.

So, we actually want more of these engineers, not less of them, uh, as an industry. And I think that's going to play out in a lot of examples, um, where it turns out that AI is going to make it a lot easier to do a particular task. And because of that, the human that sort of can help manage supervise those tasks and can help do the jobs that surround that task will actually be even more in demand. And so I think a lot of folks right now are probably being overly fearful. And the best thing they could actually do is to say, I'm just going to get really good at wielding these tools because if I am, my services are going to be highly valued in the market.

and that will be exceptionally good for their career. >> Yeah, I am in full agreement with that sentiment. My firm does test preparation for the SAT, ACT, college counseling, anything around higher education, helping students reach their goals for the future. And we have been using AI to make our teachers more effective to automate tasks in their pipeline of of the work they do where it's okay. We're going to make this automatically summarize everything you've done with this student to this point so that right before you meet with them, you get a refresher on every the all the ups and downs and struggles and triumphs with this particular student.

And it's interesting that some of those teachers though have then said are is the where's this going eventually? Is this are you looking to eventually automate us away completely where it will actually do all the teaching for us? And there is talk of that within the [snorts] industry that that's going to happen. And an example that I keep coming back to, I was observing a lesson the other day and the student or the the teacher asked the student had they done their homework and the student said yes. And I watched the teacher just he turned his head and just looked at the student. He didn't even say any words.

It was just a look. And then the [snorts] kid cracked and was like, "Well, I not really. I've been really busy. I had this going and this going and I I I mean I kind of rushed through it and I I mean I no I didn't really do it. And that little moment I just thought that's why we still need a human in the loop. And just like you said with the engineers yeah there are a lot of tasks around the job that can be automated and I think this applies across industries but that engineer their job isn't to actually code. That was just a major task in the process. And when I think about the teaching example, I go, you're there to help this younger human reach their goals for the future.

And there are a lot of things we can automate. So you focus on that piece of it because that's where the real deep value is. >> Oh, I a thousand% agree with that. Like, you know, I have a a number of teachers in my family and you spend time with them. You know, these teachers are overwork, underpaid. They had they do so many tasks beyond just simply classroom tasks. There's so much work that goes on, so much administration, so much coordination tax, so much of these things. If we merely just automated that part of the job, that would be a huge boost to helping our students get better outcomes. But you can go even further which is to say hey you now can actually introduce like worldclass intelligence to the classroom.

You know how could one human ever be able to teach these concepts to the degree that an AI could actually teach them? But what the and that then changes the role of the human. The human uh teacher can do the things you're talking about which is like hey I'm going to be an accountability partner. I'm going to be a guide. I'm going to be a coach. I'm going to push you and navigate you. I may not always be the expert at a particular topic, but I can help you navigate, you know, this learning curve that you're on um you know, really really effectively. And that's actually pretty exciting when you think about like how do we actually go prepare the next generation to to be more, you know, to push the limits of society, to be more entrepreneurial, to solve more problems for folks, to be productive members of society, all that sort of stuff.

We should want to reinvent our education system to take advantage of these new capabilities and we should want to equip our teachers to be able to use those tools very well and then also like innovate on their jobs too to be able to make it uh easier for them to to advance our you know the outcomes for our students. >> Absolutely. Absolutely. I mean, we're going deep down the teaching rabbit hole here, but I had made the argument before that a great teacher could actually teach anything through those mechanisms that you just mentioned, being the right role model. And I once worked with a student that was uh studying for physics and I do not have a great physics background whatsoever.

But it was really a matter of well okay so there is a class there is a rubric for it there are things you're supposed to learn along the way and just by being that that coach that mentor that accountability partner I was able to help this student succeed in that class and I think wow that would become 10x more true if we also have the worldclass partner sitting right there that we can go we can actually ask the physics questions to and I don't have to have that knowledge But there's still this critical role there uh to make the process successful. And whether you're a doctor or a plumber, I think that is going to be true across the board.

>> 100%. >> Um 10 years from now, what do you think we're going to look back on in 2026 and be like, "Wow, we really missed that." Or or the big surprises that might be down the line with all of this. You know, there's a saying which is we often overestimate what can happen in one year and underestimate what can happen in 10 years. And I found that to be true. And I think right now we will look back on 2026 and be shocked at how far we've come. Like we'll be like, "Wow, the world looks very, very different." and we'll probably get to the end of 2026 and say wait I like you know I I thought AI was supposed to change everything [laughter] >> right >> and it turns out like oh it's it will it's just going to take time for it to sort of work through uh the technology to be there but then also to work through society to work through these organizations to work through folks to actually have all those changes take place um and so I do think a decade from now we will look back and you know all the things we're talking about today will probably have come Drew and probably many more that we didn't even think um to to talk about.

>> I like that because there's hope in that that there there is actually time to go out and learn and it's never too late to start and amplify what you're capable of doing with these tools. [snorts] Um Wade, I know I I I hate to end it, um but I know there are many demands on your time. really enjoyed your perspective and vision for the future. If people want to find more about you online, your company, where are the best places to learn more about all you're doing? >> I'm active on LinkedIn and ex uh Wade Foster on both. Uh you can check out Zapier, zapier.com. Uh build your first agent uh and let me know what you think.

>> Absolutely. We're going to have a link in the show notes to the agent builder because I am such a huge fan of it. And yeah, thank you so much for your time, Wade. >> Yeah, thanks for having me, Nick. >> All right. >> Okay, everybody. Until next time, ask questions, don't accept the status quo, and be curious. >> The Nick Stanley Show.