Guru CEO: Not Without Human Experts

They are quite powerful and quite capable, but they know nothing about your company, your team, your people, your products, your processes, things like that. But well, happily respond to you as if they do. The person that you want the AI to get that information to is the one who won't know if there's a hallucination and therein lies the real problem. AI amplifies the human that is working with it. And all the good ways and all the bad ways, but but amplifies that human versus replacing it. We are all managers now. Five years from now, what's going to separate the companies that actually benefited from AI versus the ones that just paid for a bunch of tools and got destroyed.

Rick, welcome to the show today. Where are you coming in from? Thanks, I'm here in sunny Philadelphia. Nice. I've been for a very long time since 1999. Yeah, yeah, beautiful place. Rick, let's jump right into it. So you've built two companies that are around this problem of companies have all this data, all these systems and getting them to work together and talk together. But now that we have artificial intelligence tools, has that problem been solved? Are we on the way to solving it? I do think we're on the way to solving it. I think that in a large sense, where we are at right now is the models themselves hit a tipping point in the last five or six months.

I think objectively that's sort of, you know, everyone sort of felt that. So they are quite powerful and quite capable, but they know nothing about your company, your team, your people, your products, your processes, things like that. But we'll happily respond to you as if they do. And that is, you know, in our experience, one of the biggest sources of hallucination. So I think in a lot of ways, how we're sort of seeing this play out is companies looking to sort of transform how they work with AI is actually making this data and accuracy problem become a very prominent fixture that needs to be addressed.

People are sort of very explicitly kind of dot connecting that to our ability to deploy agents that actually work well consistently. Is this very binary? You know, do they have good company context that's accurate? Yes or no. So I think it's getting a lot of attention as a result, which is, you know, a good thing and something we kind of hoped would happen as AI got more popular. But I think that's kind of where we are today. Yeah. And I think we all feel that our personal productivity, those are that are using the tools on a daily basis has skyrocketed. I know mine has. And yet, organizational overall productivity seems to be lagging behind that and a much more difficult challenge.

Is that ultimately where your firm steps in and what you guys do? Yes, that is where Guru comes into play. And if you sort of think about AI or agents and sort of a single player or personal type use case, you know, open claw being a huge explosion of an example of an agent built around a pattern of personal use case versus multiplayer or team based agent use case. That's a lot of where this kind of comes up. And to give that a more real feel or more real example, you know, we can hook these agents into our email and into our calendar and into our personal documents, our personal notes, our personal meeting recorders with great efficacy.

And there's certainly prompting and skill design and things that need to be figured out. I'm not saying it's like, and we're done, you know, but you can judge the accuracy of your own stuff pretty easily, right? You know what's on your calendar. And AI comes back and says, you have a meeting at two o'clock today, you know, you can objectively confirm that to be true. Right. When you shift over to an organizational pattern, now you're talking about accuracy as being something that is managed in much more of a decentralized way inside company. So the way a product works, the way we handle refunds, if that's part of our business model, the way we talk about our position of product, the way we position against a competitor, on and on and on, these use cases are where you have these groups of subject matter experts inside companies.

They know their domains. They know, you know, what that answer is, what is the answer the company stands behind? That's where it gets a lot more messy, a lot more challenging. And yes, where we are sort of squarely focused in helping companies, especially in domains where there's a high consequence for inaccuracy. If you give a wrong answer, it's a very negative ramifications of the business. Maybe you're regulated or maybe you just thrive on good customer experience. If you give a bad, a bad answer or a slow answer, you know, it's like bad for business. So that that's usually where it comes into the picture.

Yeah. Along those lines, you had a post on LinkedIn recently about how there are certain companies that are really used to those restrictions. If you are a bank, a financial firm of some kind, there's a huge penalty for giving bad information. It's not just you upset the customer and there could be legal ramifications. You could go out of business very quickly. Should it be found you're giving inaccurate information? And one of the challenges with AI, especially when I'm hearing from you on an organizational level, are those hallucinations that we're also familiar with. And if we're trying to tie all this company information together, well, the person that you want the AI to get that information to is the one who won't know if there's a hallucination and therein lies the real problem.

Let's let you just run with that ramble for a second. Yeah. Oh, you nailed it. Yeah. No, you nailed it. I mean, this is kind of the team, you know, the team agent kind of world we're now in where, yes, you have, you know, maybe it's a call center or, you know, maybe it's an HR or people domain, something that's sort of quite sensitive, even even sort of like product expertise, product capability where to your point, if you misrepresent the product, very bad things can happen. What happens is, you know, you have subject when it works well, you know, you have subject matter expertise in the company that are sort of injecting their knowledge into the agent.

They're making that knowledge available less for them to consume and more for this very broad team inside the company to be able to use, right? And so I use that customer service example before. So if we just sort of stick with that one, you know, this large customer service team, they're getting customer problems. They're turning to that AI to say, help me navigate through this. And they need to build trust in that AI. And so it's very important that you understand where did that AI get those answers from? Are those answers verified by experts inside the company? Yes or no? How often are they verified by those experts?

And that's sort of the experience that that large team needs. And then the people that are managing these agents, they kind of need to sort of look at it from the inside and go, what is this AI saying to all of these people? You know, and that needs to be very clear, very transparent, very observable. Literally the questions and answers that are getting asked and that the AI is returning back and then have the ability to actually change those answers and improve them. And so, you know, there's a very intentional kind of thing that's happening there that, you know, without that, if you sort of overlook that and you just sort of start hooking the AI into document repositories and all these different places, you're going to kind of get what you get.

And there's not really going to be this sort of assertion of accuracy layer that's like actually saying, no, this is the right thing. You know, we've talked about pricing all these different places in Slack and in some meeting recordings and all these different documents. So here's pricing. It's like, well, no, there were some half-baked decisions in there, but you know, the AI doesn't know, right? And you know, say, here's what you should say. And you know, if you're not careful, you'll run with that and kind of like bad things happen. So, so yeah, that that that's a great example. I think you you call it out there.

And it says it so confidently because that's right. That's how it's been trained to operate. And the danger with that is the scale I would imagine because I've been there before where I've written a newsletter email that had one error in it. And it was disastrous because it went out to more than 10,000 people as opposed to one conversation where I might have figured out that that little error was in there and corrected for it. And now you're talking what feels like an authority for the company saying the wrong thing to however many employees you have. It could be a large company with a hundred thousand employees.

Yes. Yes. This is the power of persuasion, right? And like it's interesting. Anthropic recently did research on this too. And they said, Hey, like as these artifacts that that models are generating these like nice polished documents and things like the better and better they're getting as humans, we look at those and we're like, my God, this is so professional looking look how polished this thing is like, I'm good to go. And you just sort of let it rip. And then you're like, actually, it's got made up quotes and, you know, false information on how the pricing works and, you know, totally missed the lane on this one competitor thing.

You know, but so it's getting harder, right? Like humans have to judge. Humans really do matter in this whole world, right? But but yeah, that's exactly right. You know, you, you sort of have a one to punch here where, you know, you have, you have this sort of humans asking questions to help them do their work better. And that has a blast radius that's meaningful where, you know, one bad answer can show up in all those customer conversations. Then you have the two of the one to punch, which is agents that are starting more and more to do autonomous work. Agents need this knowledge. You know, there's two audiences of this knowledge in companies, it's employees.

It's also it's also agents, other agents being built. Well, they're going to do the same process of looking up that information, getting that context of whatever the topic of knowledge might be. And then just sort of running with it in more of an autonomous fashion. So to your point, like that's where the blast radius can be even higher because if you're not looking after this thing, it's going to keep on chugging away using bad and accurate stuff and causing, you know, causing issues, very solvable problems. But you know, just sort of need that intentionality and designed to make sure it doesn't happen.

You mentioned there the importance of the human experts involved. And I want to get into that because I thought you have a very interesting point of view on the importance of the human expert in any of these roles, whether it's the expert engineer checking the code of the program that looks beautiful on the surface, but might need to really be examined under the hood or if it's an attorney and going to that expert attorney instead of just going with the boilerplate contract that Jim and I spun up for you. Because so many people are worried right now about losing their job and especially people who are experts in their particular white color industry.

How do you see this? Where are we right now? And then how do you see this developing in the future? Yeah, I'm, I see sort of a fairly explosive job growth happening because of this world, but I think there are some fairly explainable reasons of what's going on. I mean, but to sort of start with your first question, I think that the framing that lands the best or that I have just become a firm believer in is that AI amplifies the human that is working with it and all the good ways and all the bad ways, but amplifies that human versus replacing it. And I think there's a few reasons for that. You talked about the legal example and my one of my close friends talks about how, you know, these models will generate amazing contracts.

And for any of us, non lawyers, we're sort of like, it's good to go. Right. And then you give them the lawyers and lawyers go, yes, pretty good. That's not the problem. You know, the problem is what are you really trying to do that's causing you to create this contract? And that's where the complexity and what's really happening starts to cause real problems. If you aren't really thoughtful about it, I think the same thing happens with a gente k engineering. I mean, you know, it's, it's pretty funny if you take a pause and you go, the term vibe coding came out, I don't know, fairly recently, right?

Like AI is a, AI time is a vacuum. But the person who coined it is Andrei Carpathi. Andrei Carpathi is, you know, one of the most brilliant and respected engineers around. And so when he says vibe coding and how he can talk to his AI and it is writing good code and outputting good code, he is a really good judge of what good code looks like. And where the narrative went off the rails is that a typical person can just create an application that is intended to be used in a real intensive way. I'm not talking about something that you want to get working for a prototype or even for a small user population, what I'd call a hobby type.

I'm talking about something that has to be used at a sort of big scale. That's where I think a lot of the narrative went off the rails because what we sort of forgot along the way is, you know, not that these agents can't create really great, really strong code, but humans are still involved in the loop in steering, guiding, making critical decisions at critical moments in order to keep that thing on the right path. So it's just another example of my broader point. The success mode is the human expert plus AI amplifying them. That's the collective unit that you're going to have a lot of success if you work with.

If I have a people HR matter, I don't go right to some AI thing. I go to my people leader and I know she's using AI and I want her to be. And that combination of her plus the AI is going to lead to something, you know, 10x better than if the human was on their own. That's the pattern that I think works. And that's why why I talk about the job explosion is kind of my final, you know, to like close the loop back to your point on the fear part of all of this. All of the job loss fear you're hearing about is coming from tech companies themselves. And so on one hand, people will read that and go, that's the canary in the coal mine.

They are there furthest ahead in understanding this stuff. And so they're doing it boy, just wait till, you know, just wait till the gen pop rest of the industries start going, right? And I think there's two flaws with that. Number one is there is, and this is in my term, I think Sam Altman came up with this. There's AI washing going on, absolutely 100% where companies are overstaffed or have some, you know, business level issue. And it is very easy or convenient to say, oh, we're reducing staff because we can replace those people with AI. We're sophisticated. We built systems, AI systems that can replace all these people.

We don't need it. Right. By the way, I think that era only lasted about two months though, because in the beginning, it was causing stock prices to pop. Like, whoa, look what they're doing. Now I feel like investors have, and now actually stock prices are going down when those announcements happen. So I think that's actually all going to shift, which is good because nothing good comes from that because it's not true. But also like it's only making everything worse, right? But I think the bigger and more interesting thing behind all of this is like, let's say those tech companies represent, you know, two or three percent of the businesses that are out there in the world.

And if you look historically, technical talent all wanted to go work at tech companies, right? It was the obvious career path here. If you're an engineer, you want to go apply that engineering to a tech company, right? That was such a big thing. And so what's going to happen is there's going to be this, the fastest growing job right now on LinkedIn's job boards is AI engineer. And I think that is going to continue, right? And AI engineers, one version of this job explosion that I think will happen where, you know, people are learning how to cope with a gentic tools. That is the profession that is emerging called AI engineering.

And I think there'll be one in general other ops, AI operational type roles and expertise, learning how to take this technology and help a business work better. It's just going to be like a just a huge explosion because you're going to have all these other businesses that have the ability to hire talent that previously would only manifest inside this small microcosm of tech companies. Now, you know, now, now the world's oyster, right? Because so many companies, the line gets blurred. Like what is a tech company when everybody is rethinking how their businesses work with AI, the aperture just gets like huge.

So I think it's actually a pretty exciting time more than it is like a apocalyptic time. Yeah. So how do you respond to Jack Dorsey? It wasn't that long ago said he wants to just get rid of all middle management and everyone that is still left in the company will report directly to him because AI is going to run that middle management layer. And my, I was actually thinking about this when I was listening to your interview with Wade Foster of Zapier, how the leadership directs the culture of a company. But I feel like it's the middle management layer that actually executes the culture of a company.

And when Jack Dorsey says we're going to get rid of that layer and replace it with AI, I thought, well, you're going to throw out the culture of the company by doing that. I don't know if that experiment will be successful. It's, listen, it's an experiment and we'll see. And I think they'll be a lot learned either way, for sure. My thinking, though, to answer your question is I think the thing that sort of gets missed from this conversation of the flattening the org and the exponential output of less people, there's some merit to that in theory. But the thing that I think people will learn the hard way, if they don't lean into it, is models are a aggregation of all current human thought and insight.

Right. So by definition, it is the normalization of all thought, idea and insight. And so if you give all of your company know how and the way you operate and the way you build and the way you service customers, whatever your thing is, is a business that makes you matter. If you sort of start giving all of that to a quote unquote, normalized or genera-sized model, you sort of by definition commoditizing yourself. And I think the companies that embrace this human AI partnership will be the ones where the novel thought coming out of the human brain going into the AI system will keep your business being good at the thing it's good at.

And so I just feel like that needs to be way more in the conversation than it is right now. And to your point on middle managers, people who are closer to the problem in the room with the customer talking about the fit, whatever role you want to point out, they're going to have novel unique insight and that insight is going to change over time. And I think the companies that figure out how to put that novelty back into these automation systems will thrive. I think the ones that just sort of go, what can I automate next are going to be like, yeah, now you look, feel sound and are like everyone else that's doing that too.

You're sort of like a company turned into AI slap basically. Right. Yeah, that's an interesting idea because if we think about how just all the capabilities that are there for any entrepreneur to take advantage of just from the internet, we separate that from artificial intelligence. Those opportunities and tools we know well, and I feel like one of the big changes from that is it forced people to focus down more onto what's the specific value I or this team can add. And artificial intelligence is just taking that up to another level to where yes, we want to automate everything that is not unique and really valuable.

I mean, that's what I'm hearing from you. Yes. Well, and you said culture earlier too. And I think culture, you know, like everywhere, right, there's pendulum swing, right? And we had a big culture swing for a while. And, you know, boss system was dead, employee empowerment, employee had the leverage, you know, and then and then and then coming out of zero interest area, you know, the pendulum swung back, boss system is back, right? I think we'll swing back. And I think what's going to happen, you know, culture is as important as the technology. It's as important as a strategy, the business, right?

It is the it is stood the test of time through all these things. And there's a real thing happening right now, you know, we talk a lot about how to bring AI into companies. And we always say it's 40% technology, 60% human. And this is transformational stuff, right? The printing press, you know, we don't write books by hand anymore. It's just over and people are wrapping their brains around that. And so the way leaders show up is a huge deal. And yes, you could decide to go in tomorrow and be like, FYI, we're, you know, intending to reduce this team. And, you know, it's going to be automated or die.

And, you know, you could drop all of these sort of framings on the team. That's going to cascade out. And those people are going to leave. And those people are going to be ambassadors of your company no longer employed at your company. And they're going to go to other places and they're going to have a perception of your company. And you'll either care about that or you won't. But my point of view is always like, the world is small. And, you know, the way you think about that relationship between yourself as a leader and your employee is at a very critical and sensitive moment right now. And the problem is that the employers have the leverage right now.

And so when you have that, what do you do with that? And any of the companies that really stand up and go, know this matters, we're going to be thoughtful about this. If we have to do layoffs, let's just own it and say we overstaffed or the revenue isn't where we wanted it to be just own it. It's not that it doesn't happen, right? It's how you message it and how you take ownership of it. So I do think there are these like critical culture moments right now that are going to cup the chickens will roost for sure. It's just a matter of when. Yeah. Yeah. I think of numerous examples with that. I mean Nordstrom comes to mind where it was a policy that you could return any item for any reason.

And then there was the famous case. I mean, however, many decades ago somebody brought in, I believe it was a set of used tires and said, you said I can return anything. And they accepted the return of a product they clearly didn't sell. But to me, that was magical. I said, well, that that's in such a message that's going to echo out every, I mean, I know the story, not just people in the company that we are about customer service here. That matters to us down deep. You do whatever you got to do to make that customer walk out happy. And that's a difficult one to automate away. That's real. That's a differentiator.

That's a moat. That's a value. And that's where I think culture really matters for a firm. It's a huge example of culture. It's like this was actually measured. And it was like, you know, they studied a correlation between CSAT. So how are customers grading you and the customer service they're getting from you and the employee internal employee MPS internal employee satisfaction. Unsurprisingly, they found an enormous correlation. Southwest Airlines being a pinnacle example of this where everybody thought of this brand for a very long time. It's like amazing customer experience. Guess what? The employees actually love working there and they're well treated.

Turns out the two things are highly correlated and related. Yeah. And we used to use this framing it guru. When we started building our AI stuff before the the GPT three, you know, revolution began. And we would say white gloves not pink slips. And it's very much what you're getting at, which is, hey, with the additional capability you have now because AI is taking the road work away from this person's job, you have an opportunity to wow and amaze your customers. And I can point to so many of our customers that have embraced that. And they just think about things in such a different and strategic way.

And then by the way, you look at their revenue trajectory and it sort of maps directly to that it is non coincidental. So yeah, I think your point is spot on. One of the features I like best about granola during the conversation, I can ask granola what things were previously discussed during the conversation. That means I don't have to interrupt the conversation, scroll through notes to find exactly what was said, or rely on my memory to find an important detail. And the same thing applies after the meeting. I can ask what deadline did we agree to what concerns were raised or what should be included in the follow up email?

Granola answers the actual conversation questions. And when I have several meetings organized inside a folder, I can click into that folder and then into that specific meeting to answer my questions about what's going on with the project. This is especially useful for interviews, client calls, and projects with multiple conversations. Instead of trying to remember where something was said, I go directly to the relevant meeting and just to ask for it, granola doesn't just record the conversation, makes me information inside that conversation much easier to find and use. Check out granola today with the link in the description.

You said I'm in a paraphrase here, but that you can often tell quickly whether or not an AI rollout is going to produce real ROI. And I was wondering what are some of the red flags? What are some of the green flags? If we could say that? What are those things you're looking for when somebody comes to you with something? Hey, I built this over the weekend and you get a sense of whether or not it really could be a difference maker or not? Yeah, the biggest one is if the companies organized such that they have a natural dot connect between the technology and applying it to real business problems going on inside the company.

And so the way we talk about this a lot is that a group that is working on some AI initiative could be HR customer service for wherever it is, must have the following skill sets in it. This could be in one human brain that's kind of rare, but the group, this skill set needs to exist in this group. And that is one, there needs to be a red pill thinker in the room. So this is the person who's like, I'm going to learn this stuff, I'm obsessed with learning it. Like, you know, I'm going to build my open claw, I'm going to do it right. You could call it hobbyist, but like someone who is, is I will say like, genuinely or organically curious and craving and pulling that information.

No one has to tell them like, Hey, did you do the chat GPT training? Right? Like they're sort of like they're already spending their free time doing that stuff. Yeah. Yes, that's good. That's that skill number one. Skill number two is the operator person. The operator person understands some part of the business and knows how it works and knows what's bad about it and knows where the friction is and knows how it causes a team a meaningful group of humans pain and can actually talk about that and can say things like, Hey, like our, you know, stick with customer service, our CSAT is not where it should be because our average handle time, the time it takes us to resolve issues for customers is twice as long as it should be, right?

Example like that, operator skill set. And then the final one, the final skill set, I'll sort of bucket in security or you could, you could sort of say security slash technology, but like someone who, who understands that you can't just put a given agent free will over your inbox and hope for the best, right? Like, like, like knows based on the business and how it operates, like maybe you're under certain compliance or certain, certain things like that, right? So like, you know, understands the idea of like permissions that exist in systems you're connecting with. So those three skills. So that's the, the first thing we do, you know, your greener at flag is do we see the manifestation of that, of that group?

And the red flag version of it is someone who comes in and they're talking very sort of tech first and talking about theoretical architecture or capability, meaning like, Oh, you know, which models do you use? And we'll say, why? And they'll say, Oh, why? You know, I just, I just want to make sure we can use different models. And then we'll say, okay, well, what are you trying to do? Well, you know, there's lots of different things that people are trying to do is like, okay, this is, this is going to end bad. Right? And so like, that's, that's one. Because it's not just focused on what is the need here, whether that's inside or outside the company, just the back to the basics.

What is the customer need or the team's need? And what's wrong? We're spending money on this. Yes. Why are we spending money on this? Like, I imagine it is like, all right, we just turn this machine on the super powerful large language model. And it's like, all right, this thing's on like, now what? Let's let's go burn some tokens. Like to do what? Like, what are we trying to do? Right? So, you know, we can pretty quickly kind of map that out. And, you know, one of the questions we ask when we meet with customers is we ask them to self score and AI maturity. And we explain what that means. And we say one is like, you're sort of just getting started, you're experimenting with chat tools, maybe three is you have, you know, multiple agents deployed in more than one line of business, you know, doing some amount of work autonomously for you.

And we just sort of listen. And it does manifest a song when people, when people lead with, oh, we got Gemini running and we got chat, TPT running and we got some cloth over here, you know, it's a very different answer than when someone says, oh, we recently deployed a customer facing AI that is creating these great experiences, but it's also like reducing our handle time from like minutes down to like 30, 45 seconds. We're like, okay, that is a very mature answer because that person has that person can defend the CFO conversation that is going to happen six months, nine months, 12 months of like, why are we spending so much money on this stuff?

What is it doing for us? And, you know, we just made the decision at Guru, we don't want to be on the wrong side of that conversation. If our stuff's getting used and deployed, it's for some business outcome or some or some goal. We think that's how you become bubble proof. If you're sort of locked into this business problem and not sort of like a horizontal, nondescript, you know, AI at work kind of thing that you can't really tie to to a business problem. So that's how we've sort of coached our team to say like, you know, might be tempting, but when someone shows up and says, I want to, you know, want to get me some chat for my employees and they can't really go beyond that, like, you know, you're either going to lose them now or you're going to lose them in a year, but it's not going to be a great, it's not going to be a great partnership.

Right. Now you said bubble proof. Would you would you let's just expand on that? I'm curious with that bubble is and what you think might happen there. Yeah, the bubble is in a non dramatic way. Like I don't think markets are going to collapse here, but like we're in 98 or 99, you know, and like, and like, I actually occasionally will go back into the Google Books archive and read some of the tech magazines from 98 and 99. It's actually in a kind of a nerdy way, really fun to do because you can effectively peer in to the world when the internet, a lot of people like to say like, Oh, but AI is much bigger.

It doesn't matter. Our human brains don't work that way. At the time in 98, 99, it was the biggest thing for us, right? It's the biggest thing for us and all the hype and all the fanciful things going on. And the metrics that companies were going public on were pretty far removed from revenue. It was like eyeballs, which is like, how many people come to your website? Like we were, we were like literally doing stuff like that and doing IPOs like that. So that was an actual market correction. But the analogy I like to use today is token maxing and how, you know, a lot of people are sort of unapologetically like sort of the wall, you know, putting the wall up of like who's burning the most, right?

I think it will, it will definitely end. It's actually think it might end sooner than I'm even sort of thinking, because I think people are pretty quickly realizing like, if you're metric internally, like how many tokens you can burn on stuff, like you can, with will, you can win that game. Like you can, you can make a thing, do something non productive and burn a massive amount of tokens in a very quick, but more, you know, more, the more productive version of that point is tokens are like the furthest thing from business value. They are the most generic. It would be like, if you were like, hey, how many watts of electricity did you burn in your office today?

You'd be like, I don't know. I've got like some lights on here. Yeah, there's probably watts flying really well, how many watt? I don't know. Well, you got to you got to burn more watts, man. You're not, you're not, you're not wadding, you know, it's like, what did that have to do with, you know, with how the business is going? So I that that's kind of why that's kind of why I'm calling it a little bubble proof is like, I think token is like about as extreme away from how something's making a team work better as you can get. And so I think that if that becomes the language that stays, there's going to be your reckoning of some sort of like, again, I don't, I don't want to dramatize it.

It's a market correction, like, definitely, there'll be this like, what were we thinking moment that I think will look a lot like 98, 99 behavior. Right. Well, and to bring everybody up to speed, if you're not following on tokens, so that would just be tokens are just the proxy for how much AI compute you're using to build something and run something. And then along those lines, I believe there was a memo that came out of Microsoft recently, and they said their token use had gotten so high on these open API relationships with Claude code primarily that somebody looked at it and said, Oh my goodness, a number of these people, it is more expensive to run their AI agents than to just pay that person to do the job that these agents are doing.

And so they have completely shifted what they're allowing employees to do. And that's the most extreme example I can think of just token maxing, which seems silly. Although I've been guilty of it myself, I will say there was a there's a friend of mine. We're both working on apps simultaneously. And we were texting while both working in Claude code. And it was kind of like a running joke throughout the day on who was using more tokens that day. So I can get sucked into it. You can get sucked. Of course, of course. And listen, like that actually sounds pretty fun. I think when it is, you're talking about an organization's budgeting process and how they're spending and forecasting and how they're managing cost against revenue, I think that's where it can get dangerous really quickly.

But yeah, the stuff does definitely add up really quickly. And look, like it is not lost on me in theory. One employee and we have them here at Guru and they are amazing. They can spend their salary and tokens to your point, right? And there is an answer to that that says that math works because the output, the amplification of that human is such that you can justify that math, right? So I'm not sort of shooting at the idea of this stuff getting outrageously expensive. My point is merely, what did you spend it? What did it do for you? How did it help you? And that, you know, I think I think if we just need to stay on that talk track, we need to stay on that narrative that has to continue to matter.

And we have to answer that at some point. There are good answers though, I guess is my point. It's not like it's completely reckless, but it can be if you sort of lose the plot. It was like, you know, 15, 20 years ago, there was a K-lock metric that existed for engineers where engineers were being measured and evaluated based on how many lines of code they wrote, how many thousands of lines of code they wrote and you would compare each other. Now, of course, any engineering team would laugh at that rightfully. So because more code doesn't mean better code. In fact, usually it's like quite the opposite, right?

Like that's, I think the pattern that will will happen not too long from now, you know, similarly. 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 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, to your tools can finally play nice 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, never take lunch, 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 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 writing what you want to happen. Literally, you tell chat GPT, send this file to my team in Slack, update the spreadsheet and draft an email, and Zapier plus chat GPT, figure out the right tools, do it for you. Getting started is simple. Head to the Zapier chat GPT MCP server and add the tools you want chat GPT 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, join the AI revolution and make a little automation magic of your own. Try the Zapier chat GPT integration using the link below. Yeah, well, and with Claude code in particular, it my and I am not an expert engineer by any means, but I know just just enough to be dangerous. My understanding of it is it can often solve a problem with code, but using a lot more code than say that really talented engineer can do it much more elegantly with fewer lines of code and it's tighter and it works better. And the I thought it was interesting you and Wade got into that a little bit just on the importance of humans needing to retain their authorship, especially when it comes to coding with this sort of stuff.

It's like, yes, you can use these tools to help you along with the process. But if you were going to pass it up the leadership chain and show it to someone and say, Hey, when now we're ready to deploy this thing, now suddenly you own it and you have to take responsibility for what's in there. And if there are errors, they are your errors, even if you didn't write them. And I immediately thought of, well, yeah, if you wrote a paper in Microsoft Word and there are errors in there, you don't blame the software, you wouldn't say it's were Microsoft words fault for those. It's like, no, you that's right.

Yeah, Claude did that. Yeah, yeah, totally. Our VP of engineering uses this phrase internally that goes, we are all managers now. And what he means when he says that is managers are accountable for the performance of their team. If a team is not performing well, the manager has to help that team improve their performance and is accountable for doing that. That is the exact analogy when we move into the world of managing agents is whatever output they are outputting, you as the owner manager of that agent are responsible and accountable for the work that it is doing. You cannot go, well, hey, I told this thing to do some stuff that that doesn't fly doesn't work.

That's exactly right. And I think that if you embrace that mindset, it just sort of connects to so many of the things we've talked about. It's the human AI partnership and pairing that happens. Amazing things can happen. What you can create and do amazing things. But if you sort of allow this accountability loop, even your Word doc example, if someone shares a doc with you and it's long and it's meandery and has a couple of vague points in it that are abstract and you start getting in comments, the person has to go, Oh, yeah, I'll fix that. They can't go, Oh, well, I talked to Claude and Claude did it this way.

I don't care why you showed it to me. If you think it's no good, like go work with Claude and make it that out. Don't come and show me something that doesn't work. So I think that's a really important mental model here for sure. And you're getting again at an old school business principle of a good leader in any fashion, whether you're running a team or I always like sports analogy. So if you're coaching a soccer team, the when someone makes a mistake that really good leader takes responsibility for somebody else's mistake. Yes, there was a screw up, but maybe I didn't catch it when that was made or didn't talk you through it beforehand, didn't give you the tools necessary to solve that problem when you got to it.

And that's a something that's been true for a very long time. And I don't think what I'm hearing from you is AI does not change that equation. Yeah, this is not culture. You said it earlier culture, right? Perfect example. What is the culture you want to ingrain in your company? And is it past the buck and blame the other person? Or is it a culture of accountability, vulnerability, improving for mistakes made that matters more than ever? And as we bring more AI into our work, totally, totally agree. Yeah, the work is iteration. And if you're focused on improvement and learning from every success and failure, you'll be much more likely to iterate positively towards that outcome you're trying to achieve.

Yeah, to change gears just slightly here, Rick. So the second time founder here, and we haven't we haven't covered this, but your original company, Boomi, was acquired by Dell and you worked there for a while. And it's super successful in all the ways that entrepreneurs hope their ventures will be successful. I'm curious about lessons learned from going through that experience. And now you're a second time founder and what you're applying from what you've learned from round one moving into round two. Totally. Yeah. The intentionality of culture, which we sort of talked about a few times was like a huge one for sure.

I think Boomi had has it is a great company that's continues to do great things is a great culture. But I think in some ways it was a bit accidental. And what I mean by that is, you know, I learned from a lot of people, a lot of things, but in the in the category of culture, it was like, write down your core values when your company gets to a size that you aren't just all sitting in one room working together all the time, where you can kind of like organically manifest culture. You know, for example, at Guru's when we opened our second office in San Francisco, that we say, okay, it's time to write these down.

And then build rituals around them. And so an example of a ritual, you know, our values are built into our performance review process. You know, does this person live a certain value or not? How could they improve better living a certain value? We do a town hall values in action, which we have done for over 10 years now, every town hall, we celebrate someone who is living a core value, modeling a certain behavior. Those are almost always nominated by their colleagues and co-workers, which is a really cool like virtuous cycle that happens. So I would say that was one of my biggest ones. Another one behind that is this sort of belief system around DRIs, directly responsible individuals, and the criticality of making decisions.

So on the DRI front, they're very tied together. But on the DRI front having that's at least as far as I know that that's an acronym that came from Apple. And it sort of refers to the idea that there's sort of, you know, one person responsible for specific domain in the business. And so as a DRI, your job is to make the decisions in that domain of the business, you're to seek input from others, you're to consider different options, you make the call, it is not consensus, it is not what's most popular, it is a sort of intentional decision. And then you make that decision, everybody knows it is a disagree and commit type of mindset, etc.

etc. So like that learning is like the, if there's like two enormous organizational waste moments, that's one of them, like who makes the decision on this thing? And everyone's sort of sitting around looking at each other. And then you sort of go to the worst possible answer, which is like, well, maybe we'll all try to agree together. And like, that just like never works, right? But then the second variant of that is, is speed of decision making. And that was definitely, all these are learned through just mistakes. I mean, I was 23 when I started Boomi. And so I really was like two years out of college and knew way less than what I think, things that I, versus things I actually knew.

But the speed of decision making is like the, it's better to make a wrong decision than no decision in the vast majority of times. And many have popularized the idea of one way door, two way door. I think that's a Bezos Amazon theme. I'm not sure that's where I first heard it. But the idea that, what's the one way door to a door? Very few decisions are one way door decisions. One way door decision is a decision when you walk through that door, you cannot come back. If I decide to sell Boomi to Dell, that is a one way door decision. I cannot unsell Boomi from Dell. Well, maybe I know one founder who's done that.

But I would say he's kind of a legend and not something 99% of founders have ever done. That's a one way door decision. The vast majority of decisions are two way door where you can walk through the door. Oh, I don't like it over here. You can walk back through and change. I think so many times as founders, we can parallelize ourselves and think that so many of our decisions are actually one way door decisions. And they really aren't. And if you over deliberate or you slow it down and others are waiting on you to make a decision, they're just those are the two big versions, right? Like who makes the decision?

I don't know that slows the organization down. The other one is paralysis of anyone actually making a decision that slows it down. I would say those are my big three culture, DRI, speed of decision making that I've tried to carry in and have just learned from really smart people of like, why that all matters. Yeah. Yeah. I like that door analogy. I'm surprised I haven't heard that one before. That's that's that's good because you could it's freeing. Yeah. And and would even free you to spend a little bit more time when you're like, Oh, this is a one way door moment. I'm in one of those moments.

Okay. Now I'm not going to make the quick decision. This is one where I'm going to get input from others because this one has real consequences. And the other ones go with your instincts, test, see what happens. And based on the test results, either walk back or keep running with it. Yes. Yeah. And you said test. I will I will say if there's a three A on that it is create it ties to all three, but it is a culture of making mistakes. It is the idea that if if in your company, you can empower people to make a mistake and focus on the way we talk about it, it gurus only make new mistakes, only make new mistakes.

We all make them try not to make the same one more than once. When you make a mistake, what did you learn from it? What are you going to course correct and not do it again? I think that is one of the healthiest mental spaces to be in. Many of us have likely worked in the other version of that where you are capped at the knees when you make a mistake. You are disempowered when you make a mistake. It messes with your ego, your confidence. It increases the likelihood you're going to make a mistake again the next time. And so I would say that's sort of tucked into those three, but it is like, I don't know, I think noteworthy because it's hard to do.

But when you do it, man, it's really freeing for people. Yeah. The guy I'm blanking on his first name, but Watson from IBM and also another business story from way back in the day. But there was an employee that made a huge mistake that I think that cost the company like $2 million. And this was in the 80s. So that was just a huge sum of money at that time, even for them. And they asked the CEO if he was going to fire that employee for that mistake. And he said, fire him. I just invested $2 million in him. And I thought that was such a wonderful answer along to prove the point of what you're saying.

Yeah. Yeah. Yeah. Great story. Yes, totally. Rick, five years from now, this is my last question. And then we'll talk about where to find more about Guru. Great name for a company, by the way, Tess Prep Guru's is my education firm. Yeah. So it just is excellent taste right there. Five years from now, what's going to separate the companies that actually benefited from AI versus the ones that just paid for a bunch of tools and got distracted. Great question. Yeah. I'll keep the culture part of that answer quick, but that's number one, right? It's that you've created a space for employees to learn.

You know, we internally at Guru, we call AI operations inside the company mandatory gift. So it is mandatory printing presses here. We don't write books by hand anymore. You have to change the way you work. Also, we are not going to replace you with AI. That's sort of the the the the sandwich of how we communicate it in journaling. And you like that's the obligation. That's the expectation. The reason we call it a gift is you learning these skills is much bigger than Guru. And wherever you go next, you will be better off having having known all this stuff. And so I think broadly speaking, companies that create pathways for employees to learn and grow deploy this technology aggressively to be clear and quickly lean into the technology.

And also at the same time, are focused on learning how to connect the technology to transformational work that's happening inside the business. So I think building that muscle of change management, rethinking the way you operate, rethinking the way your teams are structured, not the middle manager part so much is just like, how do you organize teams to best leverage this technology in a responsible way? You know, I think all of this stuff is going to be the make and the break because, you know, to quickly recap a point I made earlier, like the reason a company and organization exists is that it has figured out some novel expertise, capability, product, it could be anything.

It could be that your best at, you know, really low margin, you know, low cost things. And that's your thing. It could be that your premium price strategy thing. It could be that you have a novel inventory capability, right? Whatever it is for the dawn of time, organizations manifest because of some collective know how. There's not going to change in five years and you're right by definition. So again, I think the companies that do kind of what I was describing are going to continue to inject that human novelty of insight in. While recognizing we have this immensely powerful technology that can do things we've never seen before.

And that when you combine that together, you're going to continue to figure out how to grow businesses that can do things that better than or that their competitors can. And so I think that's how the next five years plays out. And like, you know, the weirdest, the weirdest part of all of this stuff is the frenetic pace at times of how it feels like this technology moving. But I actually think that'll probably settle a little bit too. There's this, this author Thomas Friedman, he writes in New York Times, he wrote this book, he's a techno very much techno optimist. And he has his chart and it says, rate of technology change is one curve, human's ability to understand that technology as it's changing, those lines diverged at some point in the 2000s.

Right. And so there's going to be some, I think pacing that'll start to happen around all of this stuff, which might make it not feel so whatever emotion you want to use anxiety fueled or stressful or dizzying or, you know, and I think they'll be a little bit of a lock in. But I think when people sort of click those ingredients of success together, I think that's what's going to separate the next generation of companies. That's well said, if anybody wants to learn more about guru, find you online, where, where do they go? I mean, I hang out mostly on LinkedIn. And so Rick Nucci, guru is get guru G E T guru.com.

And this has been really fun. Thank you for, thank you for having me. Thanks for the really great thoughtful questions. It was, it was a blast, Rick. And we will put all those links in the description. So if anybody wants a quick way to get there, just scroll down and click those. And yeah, that was fascinating, interesting, really appreciate your perspective. Thank you. Okay, everybody, until next time, ask questions. Don't accept the status quo and be curious.